publications
Select publications in reverse chronological order. See also Google Scholar and DBLP.
2026
- LLMs do not grade essays like humansComputers and Education: Artificial Intelligence, 11, pp. 100666, 2026.
Large language models have recently been proposed as tools for automated essay scoring, but their agreement with human grading remains unclear. In this work, we evaluate how LLM-generated scores compare with human grades and analyze the grading behavior of several models from the GPT and Llama families in an out-of-the-box setting, without task-specific training. Our results show that agreement between LLM and human scores remains relatively weak and varies with essay characteristics. In particular, compared to human raters, LLMs tend to assign higher scores to short or underdeveloped essays, while assigning lower scores to longer essays that contain minor grammatical or spelling errors. We also find that the scores generated by LLMs are generally consistent with the feedback they generate: essays receiving more praise tend to receive higher scores, while essays receiving more criticism tend to receive lower scores. These results suggest that LLM-generated scores and feedback follow coherent patterns but rely on signals that differ from those used by human raters, resulting in limited alignment with human grading practices. Nevertheless, our work shows that LLMs produce feedback that is consistent with their grading and that they can be reliably used in supporting essay scoring.
@article{MATHEW2026100666, author = {Jerin George Mathew and Sumayya Taher and Anindita Kundu and Denilson Barbosa}, title = {LLMs do not grade essays like humans}, journal = {Computers and Education: Artificial Intelligence}, year = {2026}, doi = {https://doi.org/10.1016/j.caeai.2026.100666}, eprint = {2603.23714}, eprinttype = {arXiv}, issn = {2666-920X}, keywords = {Automatic essay scoring, Large language models, Natural language processing, LLMs in education}, pages = {100666}, url = {https://www.sciencedirect.com/science/article/pii/S2666920X26001281}, volume = {11} } - Energy to Contest? Emotional and Multimodal Contestation of Institutional InfrastructureIn Beyond Microfoundations and Macrofoundations of Institutions: A Multi-Level Theory of Institutions, Emerald Publishing Limited, 2026.
Transforming an institution is a complex, multilevel undertaking, usually accomplished by linguistically highlighting the contradictions and hypocrisies within institutional logics, infusing economic activity with moral values, motivating contestation and collective action, and creating institutional change. Organizational research has focused on the cognitive persuasiveness of actors’ words, the emotions these words invoke, and increasingly on the emotions of accompanying images. Yet, little research examines the relative persuasiveness of words and images together in emotional and factual appeals during institutional contestation. Thus, we ask: How can multimodal messages – that combine words, images, URL links, and other elements – more effectively capture attention, invoke logical arguments and emotional appeals, and catalyze multilevel conversations. To answer this question, we gathered all the tweets, retweets, and comments associated with the Energy East Pipeline, a highly contested energy project in Canada, from the pipeline's announcement in 2013 to its protest-motivated cancelation in 2017 and beyond. We examined how actors combined images, words, hashtags, URLs, and other multimodal elements; the values and emotions evoked by these combinations; and users’ responses to these tweets. We show how taken-for-granted institutional infrastructure can become challenged through images and texts, emotional and factual claims, and other multimodal elements. We help to understand how multimodal messaging on social media shapes multilevel conversations among “relational publics” whose communications constitute and contest institutions. Last, we make a methodological contribution by analyzing multimodal texts with automated coding.
@incollection{10.1108/S0733-558X20260000104006, author = {Lefsrud, Lianne M. and Ali, Hassnain and Gutierrez Gutierrez, Candelario A. and Gehman, Joel and Barbosa, Denilson and Stroulia, Eleni}, title = {Energy to Contest? Emotional and Multimodal Contestation of Institutional Infrastructure}, booktitle = {Beyond Microfoundations and Macrofoundations of Institutions: A Multi-Level Theory of Institutions}, year = {2026}, doi = {10.1108/S0733-558X20260000104006}, editor = {Harmon, Derek and Haack, Patrick and Hannigan, Timothy R. and Etchanchu, Helen}, eprint = {https://www.emerald.com/book/chapter-pdf/11792677/s0733-558x20260000104006en.pdf}, isbn = {978-1-80592-790-7}, publisher = {Emerald Publishing Limited}, url = {https://doi.org/10.1108/S0733-558X20260000104006} } - BJPWH at COLIEE 2026 Task 1: Combining BM25 and Dense Retrieval with Year-Based Filtering for Legal Case RetrievalIn Proceedings of the Workshop on the Thirteenth International Competition on Legal Information Extraction and Entailment (COLIEE 2026), 2026.
@inproceedings{COLIEE2026, author = {Calum Kwan and Vidhi Sati and Nanribet Fwangje and Xinjia Fan and Denilson Barbosa and Randy Goebel}, title = {{BJPWH} at {COLIEE 2026 Task 1}: Combining {BM25} and Dense Retrieval with Year-Based Filtering for Legal Case Retrieval}, booktitle = {Proceedings of the Workshop on the Thirteenth International Competition on Legal Information Extraction and Entailment (COLIEE 2026)}, year = {2026}, url = {https://coliee.org/COLIEE2026/overview} }
2024
- Are Large Language Models Good Essay Graders?CoRR, 2024.
We evaluate the effectiveness of Large Language Models (LLMs) in assessing essay quality, focusing on their alignment with human grading. More precisely, we evaluate ChatGPT and Llama in the Automated Essay Scoring (AES) task, a crucial natural language processing (NLP) application in Education. We consider both zero-shot and few-shot learning and different prompting approaches. We compare the numeric grade provided by the LLMs to human rater-provided scores utilizing the ASAP dataset, a well-known benchmark for the AES task. Our research reveals that both LLMs generally assign lower scores compared to those provided by the human raters; moreover, those scores do not correlate well with those provided by the humans. In particular, ChatGPT tends to be harsher and further misaligned with human evaluations than Llama. We also experiment with a number of essay features commonly used by previous AES methods, related to length, usage of connectives and transition words, and readability metrics, including the number of spelling and grammar mistakes. We find that, generally, none of these features correlates strongly with human or LLM scores. Finally, we report results on Llama 3, which are generally better across the board, as expected. Overall, while LLMs do not seem an adequate replacement for human grading, our results are somewhat encouraging for their use as a tool to assist humans in the grading of written essays in the future.
@misc{kundu2024largelanguagemodelsgood, author = {Anindita Kundu and Denilson Barbosa}, title = {Are Large Language Models Good Essay Graders?}, journal = {CoRR}, year = {2024}, doi = {10.48550/ARXIV.2409.13120}, eprint = {2409.13120}, eprinttype = {arXiv}, volume = {abs/2409.13120} } - IRJIT: A simple, online, information retrieval approach for just-in-time software defect predictionEmpir. Softw. Eng., 29(5), pp. 131, 2024.
Just-in-Time software defect prediction (JIT-SDP) prevents the introduction of defects into the software by identifying them at commit check-in time. Current software defect prediction approaches rely on manually crafted features such as change metrics and involve expensive to train machine learning or deep learning models. These models typically involve extensive training processes that may require significant computational resources and time. These characteristics can pose challenges when attempting to update the models in real-time as new examples become available, potentially impacting their suitability for fast online defect prediction. Furthermore, the reliance on a complex underlying model makes these approaches often less explainable, which means the developers cannot understand the reasons behind models’ predictions. An approach that is not explainable might not be adopted in real-life development environments because of developers’ lack of trust in its results. To address these limitations, we propose an approach called IRJIT that employs information retrieval on source code and labels new commits as buggy or clean based on their similarity to past buggy or clean commits. IRJIT approach is online and explainable as it can learn from new data without expensive retraining, and developers can see the documents that support a prediction, providing additional context. By evaluating 10 open-source datasets in a within project setting, we show that our approach is up to 112 times faster than the state-of-the-art ML and DL approaches, offers explainability at the commit and line level, and has comparable performance to the state-of-the-art.
@article{DBLP:journals/ese/SaharBHB24, author = {Hareem Sahar and Abdul Ali Bangash and Abram Hindle and Denilson Barbosa}, title = {{IRJIT:} {A} simple, online, information retrieval approach for just-in-time software defect prediction}, journal = {Empir. Softw. Eng.}, year = {2024}, doi = {10.1007/S10664-024-10514-Z}, eprint = {2210.02435}, eprinttype = {arXiv}, number = {5}, pages = {131}, volume = {29} } - Accurate and Nuanced Open-QA Evaluation Through Textual EntailmentIn Findings of the Association for Computational Linguistics ACL 2024, pp. 2575–2587, 2024.
Open-domain question answering (Open-QA) is a common task for evaluating large language models (LLMs). However, current Open-QA evaluations are criticized for the ambiguity in questions and the lack of semantic understanding in evaluators. Complex evaluators, powered by foundation models or LLMs and pertaining to semantic equivalence, still deviate from human judgments by a large margin. We propose to study the entailment relations of answers to identify more informative and more general system answers, offering a much closer evaluation to human judgment on both NaturalQuestions and TriviaQA while being learning-free. The entailment-based evaluation we propose allows the assignment of bonus or partial marks by quantifying the inference gap between answers, enabling a nuanced ranking of answer correctness that has higher AUC than current methods.
@inproceedings{yao-barbosa-2024-accurate, author = {Yao, Peiran and Barbosa, Denilson}, title = {Accurate and Nuanced Open-{QA} Evaluation Through Textual Entailment}, booktitle = {Findings of the Association for Computational Linguistics ACL 2024}, year = {2024}, address = {Bangkok, Thailand and virtual meeting}, doi = {10.18653/v1/2024.findings-acl.151}, editor = {Ku, Lun-Wei and Martins, Andre and Srikumar, Vivek}, month = {August}, pages = {2575--2587}, publisher = {Association for Computational Linguistics} } - Semantic Graphs for Syntactic Simplification: A Revisit from the Age of LLMIn Proceedings of TextGraphs-17: Graph-based Methods for Natural Language Processing, pp. 105–115, 2024.
Symbolic sentence meaning representations, such as AMR (Abstract Meaning Representation) provide expressive and structured semantic graphs that act as intermediates that simplify downstream NLP tasks. However, the instruction-following capability of large language models (LLMs) offers a shortcut to effectively solve NLP tasks, questioning the utility of semantic graphs. Meanwhile, recent work has also shown the difficulty of using meaning representations merely as a helpful auxiliary for LLMs. We revisit the position of semantic graphs in syntactic simplification, the task of simplifying sentence structures while preserving their meaning, which requires semantic understanding, and evaluate it on a new complex and natural dataset. The AMR-based method that we propose, AMRS$^3$, demonstrates that state-of-the-art meaning representations can lead to easy-to-implement simplification methods with competitive performance and unique advantages in cost, interpretability, and generalization. With AMRS$^3$ as an anchor, we discover that syntactic simplification is a task where semantic graphs are helpful in LLM prompting. We propose AMRCoC prompting that guides LLMs to emulate graph algorithms for explicit symbolic reasoning on AMR graphs, and show its potential for improving LLM on semantic-centered tasks like syntactic simplification.
@inproceedings{yao-etal-2024-semantic, author = {Yao, Peiran and Guzhva, Kostyantyn and Barbosa, Denilson}, title = {Semantic Graphs for Syntactic Simplification: A Revisit from the Age of {LLM}}, booktitle = {Proceedings of TextGraphs-17: Graph-based Methods for Natural Language Processing}, year = {2024}, address = {Bangkok, Thailand}, doi = {10.18653/v1/2024.textgraphs-1.8}, editor = {Ustalov, Dmitry and Gao, Yanjun and Panchenko, Alexander and Tutubalina, Elena and Nikishina, Irina and Ramesh, Arti and Sakhovskiy, Andrey and Usbeck, Ricardo and Penn, Gerald and Valentino, Marco}, eprint = {2407.04067}, eprinttype = {arXiv}, month = {August}, pages = {105--115}, publisher = {Association for Computational Linguistics} } - Effective Trajectory Imputation using Simple Probabilistic Language ModelsIn 2024 25th IEEE International Conference on Mobile Data Management (MDM), pp. 51-60, 2024.
Trajectory data collected by GPS has found many critical applications. Unfortunately, most trajectory datasets have missing data due to technical problems or due to the sampling strategy used. Trajectory imputation is the task of filling in the gaps in actual trajectories by computing points that fit naturally within existing trajectories. Considering that both trajectories and natural language are essentially sequences of symbols, we explore the use of probabilistic language models for trajectory imputation. Using a grid-based representation of the space, and not considering the underlying road network, we convert trajectory points into tokens corresponding to the grid cell where they appear and train models of different sizes. We report experiments on a real dataset of over 500,000 taxi trips, showing that we can accurately fill gaps of up to 2km between GPS observations with 83% precision. These results are comparable to approaches using much more computationally demanding Large Language Models based on transformers. We discuss why transformers are overkill for the task through experiments that show that trajectory data does not exhibit very long dependencies, as is the case with natural language.
@inproceedings{10591657, author = {Mohammed, Hayat Sultan and Nascimento, Mario A. and Barbosa, Denilson}, title = {Effective Trajectory Imputation using Simple Probabilistic Language Models}, booktitle = {2024 25th IEEE International Conference on Mobile Data Management (MDM)}, year = {2024}, doi = {10.1109/MDM61037.2024.00027}, issn = {2375-0324}, keywords = {Computational modeling;Natural languages;Symbols;Transformers;Probabilistic logic;Imputation;Data models;trajectory data;trajectory imputation;language models}, month = {6}, pages = {51-60} } - Language Resources From Prominent Born-Digital Humanities Texts are Still Needed in the Age of LLMsIn Proceedings of the 4th International Conference on Natural Language Processing for Digital Humanities, pp. 85–104, 2024.
The digital humanities (DH) community fundamentally embraces the use of computerized tools for the study and creation of knowledge related to language, history, culture, and human values, in which natural language plays a prominent role. Many successful DH tools rely heavily on Natural Language Processing methods, and several efforts exist within the DH community to promote the use of newer and better tools. Nevertheless, most NLP research is driven by web corpora that are noticeably different from texts commonly found in DH artifacts, which tend to use richer language and refer to rarer entities. Thus, the near-human performance achieved by state-of-the-art NLP tools on web texts might not be achievable on DH texts. We introduce a dataset carefully created by computer scientists and digital humanists intended to serve as a reference point for the development and evaluation of NLP tools. The dataset is a subset of a born-digital textbase resulting from a prominent and ongoing experiment in digital literary history, containing thousands of multi-sentence excerpts that are suited for information extraction tasks. We fully describe the dataset and show that its language is demonstrably different than the corpora normally used in training language resources in the NLP community.
@inproceedings{hervieux-etal-2024-language, author = {Hervieux, Natalie and Yao, Peiran and Brown, Susan and Barbosa, Denilson}, title = {Language Resources From Prominent Born-Digital Humanities Texts are Still Needed in the Age of {LLM}s}, booktitle = {Proceedings of the 4th International Conference on Natural Language Processing for Digital Humanities}, year = {2024}, address = {Miami, USA}, doi = {10.18653/v1/2024.nlp4dh-1.9}, editor = {H{\"a}m{\"a}l{\"a}inen, Mika and {\"O}hman, Emily and Miyagawa, So and Alnajjar, Khalid and Bizzoni, Yuri}, month = {November}, pages = {85--104}, publisher = {Association for Computational Linguistics} } - Improving Bengali and Hindi Large Language ModelsIn Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pp. 8719–8731, 2024.
Despite being widely spoken worldwide, Bengali and Hindi are low-resource languages. The state-of-the-art in modeling such languages uses BERT and the Wordpiece tokenizer. We observed that the Wordpiece tokenizer often breaks words into meaningless tokens, failing to separate roots from affixes. Moreover, Wordpiece does not take into account fine-grained character-level information. We hypothesize that modeling fine-grained character-level information or interactions between roots and affixes helps with modeling highly inflected and morphologically complex languages such as Bengali and Hindi. We used BERT with two different tokenizers - a Unigram tokenizer and a character-level tokenizer and observed better performance. Then, we pretrained four language models accordingly - Bengali Unigram BERT, Hindi Unigram BERT, Bengali Character BERT, and Hindi Character BERT, and evaluated them for masked token detection, both in correct and erroneous settings, across many NLU tasks. We provide experimental evidence that Unigram and character-level tokenizers lead to better pretrained models for Bengali and Hindi, outperforming the previous state-of-the-art and BERT with Wordpiece vocabulary. We conduct the first study investigating the efficacy of different tokenization methods in modeling Bengali and Hindi.
@inproceedings{shahriar-barbosa-2024-improving-bengali, author = {Shahriar, Arif and Barbosa, Denilson}, title = {Improving {B}engali and {H}indi Large Language Models}, booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)}, year = {2024}, address = {Torino, Italia}, editor = {Calzolari, Nicoletta and Kan, Min-Yen and Hoste, Veronique and Lenci, Alessandro and Sakti, Sakriani and Xue, Nianwen}, month = {May}, pages = {8719--8731}, publisher = {ELRA and ICCL} } - Instant Messaging Platforms News Multi-Task Classification for Stance, Sentiment, and Discrimination DetectionIn Proceedings of the Third Ukrainian Natural Language Processing Workshop (UNLP) @ LREC-COLING 2024, pp. 30–40, 2024.
In the digital age, geopolitical events frequently catalyze discussions among global web users. Platforms such as social networks and messaging applications serve as vital means for information spreading and acquisition. The Russian aggression against Ukraine has notably intensified online discourse on the matter, drawing a significant audience eager for real-time updates. This surge in online activity inevitably results in the proliferation of content, some of which may be unreliable or manipulative. Given this context, the identification of such content with information distortion is imperative to mitigate bias and promote fairness. However, this task presents considerable challenges, primarily due to the lack of sophisticated language models capable of understanding the nuances and context of texts in low-resource languages, and the scarcity of well-annotated datasets for training such models. To address these gaps, we introduce the TRWU dataset - a meticulously annotated collection of Telegram news about the Russian war in Ukraine gathered starting from January 1, 2022. This paper outlines our methodology for semantic analysis and classification of these messages, aiming to ascertain their bias. Such an approach enhances our ability to detect manipulative and destructive content. Through descriptive statistical analysis, we explore deviations in message sentiment, stance, and metadata across different types of channels and levels of content creation activity. Our findings indicate a predominance of negative sentiment within the dataset. Additionally, our research elucidates distinct differences in the linguistic choices and phraseology among channels, based on their stance towards the war. This study contributes to the broader effort of understanding the spread and mitigating the impact of biased and manipulative content in digital communications.
@inproceedings{ustyianovych-barbosa-2024-instant, author = {Ustyianovych, Taras and Barbosa, Denilson}, title = {Instant Messaging Platforms News Multi-Task Classification for Stance, Sentiment, and Discrimination Detection}, booktitle = {Proceedings of the Third Ukrainian Natural Language Processing Workshop (UNLP) @ LREC-COLING 2024}, year = {2024}, address = {Torino, Italia}, editor = {Romanyshyn, Mariana and Romanyshyn, Nataliia and Hlybovets, Andrii and Ignatenko, Oleksii}, month = {May}, pages = {30--40}, publisher = {ELRA and ICCL} } - A Bayesian Approach Towards Crowdsourcing the Truths from LLMsIn NeurIPS 2024 Workshop on Bayesian Decision-making and Uncertainty, 2024.
Concerns persist over the trustworthiness of large language models (LLMs) due to the generation of plausible but incorrect information, known as hallucination. Existing approaches focus on identifying false answers or improving correctness by sampling responses from a single LLM. However, querying multiple LLMs, which exhibit complementary strengths, remains largely unexplored. In this work, we propose a Bayesian crowdsourcing approach towards aggregating multiple answers from multiple LLMs and quantifying their uncertainty. Extending the Dawid-Skene model, we treat LLMs as annotators, using their answer probabilities as noisy observations of truthfulness and modeling semantic relations between answers in the covariance structure, and jointly learn about LLM’s reliability and calibration as parameters. Validated across three open-domain question answering dataset, results show that our approach outperforms existing statistical or agentic methods in abstaining from false answers and identifying truthful ones, offering a robust, scalable solution for uncertainty quantification and truth discovery in LLM outputs.
@inproceedings{yao2024a, author = {Peiran Yao and Jerin George Mathew and Shehraj Singh and Donatella Firmani and Denilson Barbosa}, title = {A Bayesian Approach Towards Crowdsourcing the Truths from {LLM}s}, booktitle = {NeurIPS 2024 Workshop on Bayesian Decision-making and Uncertainty}, year = {2024} } - Orlando Information Extraction Dataset: Language Resources From Prominent Born-Digital Humanities TextsDataset, Borealis, 2024.
An annotated dataset created by computer scientists and digital humanists intended to serve as a reference point for NLP researchers. The dataset is derived from text of the Orlando Project, and contains thousands of multi-sentence excerpts that are suited for information extraction tasks such as entity linking, and relation extraction. As a whole, the dataset can be used for fine-tuning and benchmarking large language models. The dataset is further described in Language Resources From Prominent Born-Digital Humanities Texts are Still Needed in the Age of LLMs (Hervieux et al., NLP4DH 2024).
@data{SP3/RCVANO_2024, author = {Hervieux, Natalie and Yao, Peiran and Brown, Susan and Barbosa, Denilson}, title = {{Orlando Information Extraction Dataset: Language Resources From Prominent Born-Digital Humanities Texts}}, year = {2024}, doi = {10.5683/SP3/RCVANO}, publisher = {Borealis}, url = {https://doi.org/10.5683/SP3/RCVANO}, version = {V1} }
2023
- NLP Workbench: Efficient and Extensible Integration of State-of-the-art Text Mining ToolsIn Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations, pp. 18–26, 2023.
NLP Workbench is a web-based platform for text mining that allows non-expert users to obtain semantic understanding of large-scale corpora using state-of-the-art text mining models. The platform is built upon latest pre-trained models and open source systems from academia that provide semantic analysis functionalities, including but not limited to entity linking, sentiment analysis, semantic parsing, and relation extraction. Its extensible design enables researchers and developers to smoothly replace an existing model or integrate a new one. To improve efficiency, we employ a microservice architecture that facilitates allocation of acceleration hardware and parallelization of computation. This paper presents the architecture of NLP Workbench and discusses the challenges we faced in designing it. We also discuss diverse use cases of NLP Work- bench and the benefits of using it over other approaches. The platform is under active devel- opment, with its source code released under the MIT license. A website and a short video demonstrating our platform are also available.
@inproceedings{yao-etal-2023-nlp, author = {Yao, Peiran and Kosmajac, Matej and Waheed, Abeer and Guzhva, Kostyantyn and Hervieux, Natalie and Barbosa, Denilson}, title = {{NLP} Workbench: Efficient and Extensible Integration of State-of-the-art Text Mining Tools}, booktitle = {Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics: System Demonstrations}, year = {2023}, address = {Dubrovnik, Croatia}, doi = {10.18653/v1/2023.eacl-demo.3}, eprint = {2303.01410}, eprinttype = {arXiv}, month = {May}, pages = {18--26}, publisher = {Association for Computational Linguistics} } - Relational Extraction on Wikipedia Tables using Convolutional and Memory NetworksCoRR, abs/2307.05827, 2023.
Relation extraction (RE) is the task of extracting relations between entities in text. Most RE methods extract relations from free-form running text and leave out other rich data sources, such as tables. We explore RE from the perspective of applying neural methods on tabularly organized data. We introduce a new model consisting of Convolutional Neural Network (CNN) and Bidirectional-Long Short Term Memory (BiLSTM) network to encode entities and learn dependencies among them, respectively. We evaluate our model on a large and recent dataset and compare results with previous neural methods. Experimental results show that our model consistently outperforms the previous model for the task of relation extraction on tabular data. We perform comprehensive error analyses and ablation study to show the contribution of various components of our model. Finally, we discuss the usefulness and trade-offs of our approach, and provide suggestions for fostering further research.
@article{DBLP:journals/corr/abs-2307-05827, author = {Arif Shahriar and Rohan Saha and Denilson Barbosa}, title = {Relational Extraction on Wikipedia Tables using Convolutional and Memory Networks}, journal = {CoRR}, year = {2023}, doi = {10.48550/arXiv.2307.05827}, eprint = {2307.05827}, eprinttype = {arXiv}, volume = {abs/2307.05827} }
2022
- WordTies: Measuring Word Associations in Language Models via Constrained SamplingIn Findings of the Association for Computational Linguistics: EMNLP 2022, pp. 5959–5970, 2022.
Word associations are widely used in psychology to provide insights on how humans perceive and understand concepts. Comparing word associations in language models (LMs) to those generated by human subjects can serve as a proxy to uncover embedded lexical and commonsense knowledge in language models. While much helpful work has been done applying direct metrics, such as cosine similarity, to help understand latent spaces, these metrics are symmetric, while human word associativity is asymmetric. We propose WordTies, an algorithm based on constrained sampling from LMs, which allows an asymmetric measurement of associated words, given a cue word as the input. Comparing to existing methods, word associations found by this method share more overlap with associations provided by humans, and observe the asymmetric property of human associations. To examine possible reasons behind associations, we analyze the knowledge and reasoning behind the word pairings as they are linked to lexical and commonsense knowledge graphs.When the knowledge about the nature of the word pairings is combined with a probability that the LM has learned that information, we have a new way to examine what information is captured in LMs.
@inproceedings{yao-etal-2022-wordties, author = {Yao, Peiran and Renwick, Tobias and Barbosa, Denilson}, title = {{W}ord{T}ies: Measuring Word Associations in Language Models via Constrained Sampling}, booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2022}, year = {2022}, address = {Abu Dhabi, United Arab Emirates}, doi = {10.18653/v1/2022.findings-emnlp.440}, month = {December}, pages = {5959--5970}, publisher = {Association for Computational Linguistics} } - Cree Corpus: A Collection of nêhiyawêwin ResourcesIn Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 6354–6364, 2022.
Plains Cree (nêhiyawêwin) is an Indigenous language that is spoken in Canada and the USA. It is the most widely spoken dialect of Cree and a morphologically complex language that is polysynthetic, highly inflective, and agglutinative. It is an extremely low resource language, with no existing corpus that is both available and prepared for supporting the development of language technologies. To support nêhiyawêwin revitalization and preservation, we developed a corpus covering diverse genres, time periods, and texts for a variety of intended audiences. The data has been verified and cleaned; it is ready for use in developing language technologies for nêhiyawêwin. The corpus includes the corresponding English phrases or audio files where available. We demonstrate the utility of the corpus through its community use and its use to build language technologies that can provide the types of support that community members have expressed are desirable. The corpus is available for public use.
@inproceedings{teodorescu-etal-2022-cree, author = {Teodorescu, Daniela and Matalski, Josie and Lothian, Delaney and Barbosa, Denilson and Demmans Epp, Carrie}, title = {{C}ree Corpus: A Collection of n{\^e}hiyaw{\^e}win Resources}, booktitle = {Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)}, year = {2022}, address = {Dublin, Ireland}, doi = {10.18653/v1/2022.acl-long.440}, month = {May}, pages = {6354--6364}, publisher = {Association for Computational Linguistics} } - HOPLoP: Multi-Hop Link Prediction over Knowledge Graph EmbeddingsWorld Wide Web, 25(2), pp. 1037–1065, 2022.
Large-scale Knowledge Graphs (KGs) support applications such as Web search and personal assistants and provide training data for numerous Natural Language Processing tasks. Nevertheless, building KGs with high accuracy and domain coverage remains difficult, and neither manual nor automatic efforts are up to par. Link Prediction (LP) is one of many tasks aimed at addressing this problem. Its goal is to find missing links between entities in the KG based on structural by exploiting regularities in the graph structure. Recent years have seen two approaches emerge: using KG embeddings, and modelling complex relations by exploiting correlations between individual links and longer paths connecting the same pair of entities. For the latter, state-of-the-art methods traverse the KG itself and are hampered both by incompleteness and skewed degree distributions found in most KGs, resulting in some entities being overly represented in the training set leading to poor generalization. We present HOPLoP: an efficient and effective multi-hop LP meta method that performs the equivalent to path traversals on the KG embedding space instead of the KG itself, marrying both ideas. We show how to train and tune our method with different underlying KG embeddings, and report on experiments on many benchmarks, showing both that HOPLoP improves each LP method on its own and that it consistently outperforms the previous state-of-the-art by a good margin. Finally, we describe a way to interpret paths generated by HOPLoP when used with TransE.
@article{RanganathanB22, author = {Ranganathan, Varun and Barbosa, Denilson}, title = {HOPLoP: Multi-Hop Link Prediction over Knowledge Graph Embeddings}, journal = {World Wide Web}, year = {2022}, address = {USA}, doi = {10.1007/s11280-021-00972-6}, issn = {1386-145X}, keywords = {Link Prediction, Knowledge Graph Embeddings, Multi-hop reasoning}, month = {3}, number = {2}, numpages = {29}, pages = {1037–1065}, publisher = {Kluwer Academic Publishers}, volume = {25} } - The Effect of Fine-tuned Word Embedding Techniques on the Accuracy of Automated Essay Scoring Systems Using Neural NetworksJournal of Applied Testing Technology, 23(0), pp. 21–29, 2022.
Automated Essay Scoring (AES) using neural networks has helped increase the accuracy and efficiency of scoring students’ written tasks. Generally, the improved accuracy of neural network approaches has been attributed to the use of modern word embedding techniques. However, which word embedding techniques produce higher accuracy in AES systems with neural networks is still unclear. In addition, the importance of fine-tuned word embedding techniques on the accuracy of the AES systems is not justified yet. This study investigates the effect of fine-tuned modern word embedding techniques, including pretrained GloVe and Word2Vec, on the accuracy of a deep learning AES model using a Long-Short Term Memory (LSTM) network. The dataset used in this study consisted of 12,978 essays introduced in the 2012 Automated Scoring Assessment Prize (ASAP) competition. Results show that fine-tuned word embedding techniques could significantly improve the accuracy of the AES (QWK= 0.79) compared with the baseline model without pretrained embeddings (QWK = 0.73). Moreover, when used in AES, the pre-trained GloVe word embedding (QWK= 0.79) outperformed Word2Vec (QWK = 0.77). The results of this study can guide future AES studies in selecting more appropriate word representations and how to fine-tune the word embedding techniques for scoring-related tasks.
@article{JATT2022, author = {Tahereh Firoozi and Okan Bulut and Carrie Demmans Epp and Ali Naeimabadi and Denilson Barbosa}, title = {The Effect of Fine-tuned Word Embedding Techniques on the Accuracy of Automated Essay Scoring Systems Using Neural Networks}, journal = {Journal of Applied Testing Technology}, year = {2022}, keywords = {Automated Essay Scoring, Glove Embedding, Neural Networks, Word Embeddings, Word2Vec}, number = {0}, pages = {21–29}, volume = {23} } - Energy to contest? Emotional and multimodal contestation of energy marketsIn Proceedings of the 38th EGOS Colloquium, 2022.
@inproceedings{AI4BuzEGOS2022, author = {Lianne M. Lefsrud and Candelario A. Gutierrez Gutierrez and Eleni Stroulia and Joel Gehman and Denilson Barbosa}, title = {Energy to contest? Emotional and multimodal contestation of energy markets}, booktitle = {Proceedings of the 38th EGOS Colloquium}, year = {2022}, publisher = { European Group for Organizational Studies} } - Web Services for Voyant: LINCS, Voyant and NSSIn Proceedings of DIGITAL HUMANITIES 2022, 2022.
@inproceedings{DH2022, author = { Geoffrey Rockwell and Natalie Hervieux and Huma Zafar and Kaylin Land and Andrew MacDonald and Denilson Barbosa and Luciano Frizzera and Mihaela Ilovan and Susan Brown}, title = {Web Services for Voyant: LINCS, Voyant and NSS}, booktitle = {Proceedings of DIGITAL HUMANITIES 2022}, year = {2022}, publisher = {Carnegie Mellon University Press} }
2021
- Knowledge Graph Embedding for Link Prediction: A Comparative AnalysisACM Trans. Knowl. Discov. Data, 15(2), 2021.
Knowledge Graphs (KGs) have found many applications in industrial and in academic settings, which in turn, have motivated considerable research efforts towards large-scale information extraction from a variety of sources. Despite such efforts, it is well known that even the largest KGs suffer from incompleteness; Link Prediction (LP) techniques address this issue by identifying missing facts among entities already in the KG. Among the recent LP techniques, those based on KG embeddings have achieved very promising performance in some benchmarks. Despite the fast-growing literature on the subject, insufficient attention has been paid to the effect of the design choices in those methods. Moreover, the standard practice in this area is to report accuracy by aggregating over a large number of test facts in which some entities are vastly more represented than others; this allows LP methods to exhibit good results by just attending to structural properties that include such entities, while ignoring the remaining majority of the KG. This analysis provides a comprehensive comparison of embedding-based LP methods, extending the dimensions of analysis beyond what is commonly available in the literature. We experimentally compare the effectiveness and efficiency of 18 state-of-the-art methods, consider a rule-based baseline, and report detailed analysis over the most popular benchmarks in the literature.
@article{RossiBFMM21, author = {Rossi, Andrea and Barbosa, Denilson and Firmani, Donatella and Matinata, Antonio and Merialdo, Paolo}, title = {Knowledge Graph Embedding for Link Prediction: A Comparative Analysis}, journal = {ACM Trans. Knowl. Discov. Data}, year = {2021}, address = {New York, NY, USA}, articleno = {14}, doi = {10.1145/3424672}, eprint = {2002.00819}, eprinttype = {arXiv}, issn = {1556-4681}, keywords = {knowledge graph embeddings, link prediction, Knowledge graphs, comparative analysis}, month = {1}, number = {2}, numpages = {49}, publisher = {Association for Computing Machinery}, volume = {15} } - Typing Errors in Factual Knowledge Graphs: Severity and Possible Ways OutIn WWW '21: The Web Conference 2021, Virtual Event / Ljubljana, Slovenia, April 19-23, 2021, pp. 3305–3313, 2021.
Large-scale factual knowledge graphs (KGs) such as DBpedia and Wikidata are essential to many popular downstream tasks and are also widely used by various research communities as training and/or benchmarking data. Despite their immense success and utility, these KGs are surprisingly noisy. In this study, we investigate the quality of these KGs, where the typing error rate is estimated to be 27% for coarse-grained types on average, and even 73% for certain fine-grained types. In pursuit of solutions, we propose an active typing error detection algorithm that maximizes the utilization of both gold and noisy labels. We also comprehensively discuss and compare the state-of-the-art in unsupervised, semi-supervised, and supervised paradigms to deal with typing errors in factual KGs. The outcomes of this study provide guidelines for researchers to use noisy factual KGs. To help practitioners deploy the techniques and conduct further research, we published our code and data 1.
@inproceedings{YaoB21, author = {Peiran Yao and Denilson Barbosa}, title = {Typing Errors in Factual Knowledge Graphs: Severity and Possible Ways Out}, booktitle = {{WWW} '21: The Web Conference 2021, Virtual Event / Ljubljana, Slovenia, April 19-23, 2021}, year = {2021}, doi = {10.1145/3442381.3449977}, editor = {Jure Leskovec and Marko Grobelnik and Marc Najork and Jie Tang and Leila Zia}, eprint = {2102.02307}, eprinttype = {arXiv}, pages = {3305--3313}, publisher = {{ACM} / {IW3C2}} } - Detection and Identification of Obfuscated Obscene Language with Character Level TransformersIn Proceedings of the 34th Canadian Conference on Artificial Intelligence, Canadian AI 2021, pp. 8 pp, 2021.
We present O-Norm (Offense Normalizer), a system for normalizing offensive words and phrases by reversing user-created obfuscations. O-Norm is intended to be a preprocessing tool for pipelines that run on user-created text, particularly those tasks where obscenities may be of significant value such as sentiment, toxicity, and abuse detection. O-Norm is constructed with a purely generated, context-free dataset derived from a curse dictionary. The generative dataset allows O-Norm to be flexible to the addition of new words as language adapts, and also allows it to be readily retrained in other Latin alphabet-based languages as it does not require manual annotations to train. O-Norm is based on a character-level, transformer network which attempts de-obfuscation only on out-of-vocabulary (OOV) tokens. In an 80/20 train-test split, O-Norm achieves an F1 score of 89.6% over 141 curses in a generated dataset with 2.16 million unique training points. An inspection of O-Norm’s output on a sample of social media posts from Kaggle’s Jigsaw corpus reveals accuracy of 95.7% on de-obfuscating transformations in toxic user-created text.
@inproceedings{RenwickBarbosa21, author = {Renwick, Tobias and Barbosa, Denilson}, title = {Detection and Identification of Obfuscated Obscene Language with Character Level Transformers}, booktitle = {Proceedings of the 34th Canadian Conference on Artificial Intelligence, Canadian {AI} 2021}, year = {2021}, doi = {10.21428/594757db.cd61e1d6}, pages = {8 pp} } - Analyzing and visualizing Twitter conversationsIn CASCON '21: Proceedings of the 31st Annual International Conference on Computer Science and Software Engineering, Toronto, Ontario, Canada, November 22 - 25, 2021, pp. 4–13, 2021.
@inproceedings{GutierrezWPGLBS21, author = {Candelario A. Gutierrez Gutierrez and Andrea M. Whittaker and Katherine Mae Patenio and Joel Gehman and Lianne M. Lefsrud and Denilson Barbosa and Eleni Stroulia}, title = {Analyzing and visualizing Twitter conversations}, booktitle = {{CASCON} '21: Proceedings of the 31st Annual International Conference on Computer Science and Software Engineering, Toronto, Ontario, Canada, November 22 - 25, 2021}, year = {2021}, doi = {10.5555/3507788.3507791}, pages = {4--13}, publisher = {{ACM}} }
2020
- Neural Relation Extraction on Wikipedia Tables for Augmenting Knowledge GraphsIn Proceedings of the 29th ACM International Conference on Information & Knowledge Management, pp. 2133–2136, 2020.
Knowledge Graph Augmentation is the task of adding missing facts to an incomplete knowledge graph to improve its effectiveness in applications such as web search and question answering. State-of-the-art methods rely on information extraction from running text, leaving rich sources of facts such as tables behind. We help close this gap with a neural method that uses contextual information surrounding a table in a Wikipedia article to extract relations between entities appearing in the same row of a table or between the entity of said article and entities appearing in the table. We trained and tested our method on a much larger dataset compared to previous work which we have made public and observed experimentally that our method is very promising for the task.
@inproceedings{cikm2020, author = {Macdonald, Erin and Barbosa, Denilson}, title = {Neural Relation Extraction on Wikipedia Tables for Augmenting Knowledge Graphs}, booktitle = {Proceedings of the 29th ACM International Conference on Information {\&} Knowledge Management}, year = {2020}, address = {New York, NY, USA}, doi = {10.1145/3340531.3412164}, isbn = {9781450368599}, numpages = {4}, pages = {2133–2136}, publisher = {Association for Computing Machinery} } - Understanding DevOps Education with Grounded TheoryIn Proceedings of the 41st International Conference on Software Engineering: Software Engineering Education and Training, 2020.
DevOps stands for Development-Operations. It arises from the IT industry as a movement aligning development and operations teams. DevOps is broadly recognized as an IT standard, and there is high demand for DevOps practitioners in industry. Since ACM & IEEE suggest that undergraduate computer science curricula “must adequately prepare [students] for the workforce”, we studied whether undergraduates acquired adequate DevOps skills to fulfill the demand for DevOps practitioners in industry. We employed Grounded Theory (GT), a social science qualitative research methodology, to study DevOps education from academic and industrial perspectives. In academia, academics were not motivated to learn or adopt DevOps, and we did not find strong evidence of academics teaching DevOps. Academics need incentives to adopt DevOps, in order to stimulate interest in teaching DevOps. In industry, DevOps practitioners lack clearly defined roles and responsibilities, for the DevOps topic is diverse and growing too fast. Therefore, practitioners can only learn DevOps through hands-on working experience. As a result, academic institutions should provide fundamental DevOps education (in culture, procedure, and technology) to prepare students for their future DevOps advancement in industry. Based on our findings, we proposed five groups of future studies to advance DevOps education in academia.
@inproceedings{icse2020, author = {Candy Pang and Abram Hindle and Denilson Barbosa}, title = {Understanding DevOps Education with Grounded Theory}, booktitle = {Proceedings of the 41st International Conference on Software Engineering: Software Engineering Education and Training}, year = {2020}, doi = {10.1145/3377814.3381711} } - GrCluster: a score function to model hierarchy in knowledge graph embeddingsIn SAC '20: The 35th ACM/SIGAPP Symposium on Applied Computing, online event, [Brno, Czech Republic], March 30 - April 3, 2020, pp. 964–971, 2020.
Low-dimensional embeddings for knowledge graph entities and relations help preserve their latent semantics while enabling computation efficiency. These embeddings are often used to perform tasks such as knowledge graph completion, question answering and inference. Knowledge graph embedding methods aid in the representation of entities and relationships of a knowledge graph in continuous vector spaces. However, most existing techniques ignore the inherent hierarchical structure of entities present in the knowledge graph, defined by ontological relationships between entity types. This paper introduces a novel score function called GrCluster that helps fill that gap. GrCluster is a simple, intuitive and efficient scoring function that incorporates the entity hierarchical correlation into existing knowledge graph embeddings. The effectiveness of GrCluster is demonstrated by integrating it into several well known embedding models. The experimental results show consistent improvements across metrics and embedding models for the tasks of entity prediction and triplet classification.
@inproceedings{sac_RanganathanSMSB20, author = {Varun Ranganathan and Siddharth Suresh and Yash Mathur and Natarajan Subramanyam and Denilson Barbosa}, title = {GrCluster: a score function to model hierarchy in knowledge graph embeddings}, booktitle = {{SAC} '20: The 35th {ACM/SIGAPP} Symposium on Applied Computing, online event, [Brno, Czech Republic], March 30 - April 3, 2020}, year = {2020}, doi = {10.1145/3341105.3373978}, editor = {Chih{-}Cheng Hung and Tom{\'{a}}s Cern{\'{y}} and Dongwan Shin and Alessio Bechini}, pages = {964--971}, publisher = {{ACM}} }
2019
- KnowledgeNet: A Benchmark Dataset for Knowledge Base PopulationIn Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pp. 749–758, 2019.
KnowledgeNet is a benchmark dataset for the task of automatically populating a knowledge base (Wikidata) with facts expressed in natural language text on the web. KnowledgeNet provides text exhaustively annotated with facts, thus enabling the holistic end-to-end evaluation of knowledge base population systems as a whole, unlike previous benchmarks that are more suitable for the evaluation of individual subcomponents (e.g., entity linking, relation extraction). We discuss five baseline approaches, where the best approach achieves an F1 score of 0.50, significantly outperforming a traditional approach by 79% (0.28). However, our best baseline is far from reaching human performance (0.82), indicating our dataset is challenging. The KnowledgeNet dataset and baselines are available at https://github.com/diffbot/knowledge-net
@inproceedings{emnlp2019_KnowledgeNet, author = {Mesquita, Filipe and Cannaviccio, Matteo and Schmidek, Jordan and Mirza, Paramita and Barbosa, Denilson}, title = {{K}nowledge{N}et: A Benchmark Dataset for Knowledge Base Population}, booktitle = {Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)}, year = {2019}, address = {Hong Kong, China}, doi = {10.18653/v1/D19-1069}, month = {November}, pages = {749--758}, publisher = {Association for Computational Linguistics} } - Detecting and Correcting Typing Errors in DBpediaIn Proceedings of the 1st International Workshop on Challenges and Experiences from Data Integration to Knowledge Graphs co-located with the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD 2019), Anchorage, Alaska, August 5, 2019, 2019.
@inproceedings{kdd_CaminhasCHB19, author = {Daniel Caminhas and Daniel Cones and Natalie Hervieux and Denilson Barbosa}, title = {Detecting and Correcting Typing Errors in DBpedia}, booktitle = {Proceedings of the 1st International Workshop on Challenges and Experiences from Data Integration to Knowledge Graphs co-located with the 25th {ACM} {SIGKDD} International Conference on Knowledge Discovery {\&} Data Mining {(KDD} 2019), Anchorage, Alaska, August 5, 2019}, year = {2019}, editor = {Donatella Firmani and Valter Crescenzi and Andrea De Angelis and Xin Luna Dong and Maurizio Mazzei and Paolo Merialdo and Divesh Srivastava}, publisher = {CEUR-WS.org}, series = {{CEUR} Workshop Proceedings}, volume = {2512} } - Connecting Language and Knowledge with Heterogeneous Representations for Neural Relation ExtractionIn Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), pp. 3201–3206, 2019.
Knowledge Bases (KBs) require constant updating to reflect changes to the world they represent. For general purpose KBs, this is often done through Relation Extraction (RE), the task of predicting KB relations expressed in text mentioning entities known to the KB. One way to improve RE is to use KB Embeddings (KBE) for link prediction. However, despite clear connections between RE and KBE, little has been done toward properly unifying these models systematically. We help close the gap with a framework that unifies the learning of RE and KBE models leading to significant improvements over the state-of-the-art in RE. The code is available at https://github.com/billy-inn/HRERE.
@inproceedings{xu-barbosa-2019-connecting, author = {Xu, Peng and Barbosa, Denilson}, title = {Connecting Language and Knowledge with Heterogeneous Representations for Neural Relation Extraction}, booktitle = {Proceedings of the 2019 Conference of the North {A}merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers)}, year = {2019}, address = {Minneapolis, Minnesota}, doi = {10.18653/v1/n19-1323}, eprint = {1903.10126}, eprinttype = {arXiv}, month = {June}, pages = {3201--3206}, publisher = {Association for Computational Linguistics} }
2018
- Neural Fine-Grained Entity Type Classification with Hierarchy-Aware LossIn Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pp. 16–25, 2018.
The task of Fine-grained Entity Type Classification (FETC) consists of assigning types from a hierarchy to entity mentions in text. Existing methods rely on distant supervision and are thus susceptible to noisy labels that can be out-of-context or overly-specific for the training sentence. Previous methods that attempt to address these issues do so with heuristics or with the help of hand-crafted features. Instead, we propose an end-to-end solution with a neural network model that uses a variant of cross-entropy loss function to handle out-of-context labels, and hierarchical loss normalization to cope with overly-specific ones. Also, previous work solve FETC a multi-label classification followed by ad-hoc post-processing. In contrast, our solution is more elegant: we use public word embeddings to train a single-label that jointly learns representations for entity mentions and their context. We show experimentally that our approach is robust against noise and consistently outperforms the state-of-the-art on established benchmarks for the task.
@inproceedings{xu-barbosa-2018-neural, author = {Xu, Peng and Barbosa, Denilson}, title = {Neural Fine-Grained Entity Type Classification with Hierarchy-Aware Loss}, booktitle = {Proceedings of the 2018 Conference of the North {A}merican Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers)}, year = {2018}, address = {New Orleans, Louisiana}, doi = {10.18653/v1/N18-1002}, eprint = {1803.03378}, eprinttype = {arXiv}, month = {June}, pages = {16--25}, publisher = {Association for Computational Linguistics} } - Investigations on Knowledge Base Embedding for Relation Prediction and ExtractionCoRR, abs/1802.02114, 2018.
We report an evaluation of the effectiveness of the existing knowledge base embedding models for relation prediction and for relation extraction on a wide range of benchmarks. We also describe a new benchmark, which is much larger and complex than previous ones, which we introduce to help validate the effectiveness of both tasks. The results demonstrate that knowledge base embedding models are generally effective for relation prediction but unable to give improvements for the state-of-art neural relation extraction model with the existing strategies, while pointing limitations of existing methods.
@article{Corr_abs-1802-02114, author = {Peng Xu and Denilson Barbosa}, title = {Investigations on Knowledge Base Embedding for Relation Prediction and Extraction}, journal = {CoRR}, year = {2018}, archiveprefix = {arXiv}, doi = {10.48550/arxiv.1802.02114}, eprint = {1802.02114}, volume = {abs/1802.02114} } - Leveraging Wikipedia Table Schemas for Knowledge Graph AugmentationIn Proceedings of the 21st International Workshop on Web and Databases, Houston, TX, USA, June 10th, 2018, pp. 5:1-5:6, 2018.
General solutions to augment Knowledge Graphs (KGs) with facts extracted from Web tables aim to associate pairs of columns from the table with a KG relation based on the matches between pairs of entities in the table and facts in the KG. These approaches suffer from intrinsic limitations due to the incompleteness of the KGs. In this paper we investigate an alternative solution, which leverages the patterns that occur on the schemas of a large corpus of Wikipedia tables. Our experimental evaluation, which used DBpedia as reference KG, demonstrates the advantages of our approach over state-of-the-art solutions and reveals that we can extract more than 1.7M of facts with an estimated accuracy of 0.81 even from tables that do not expose any fact on the KG.
@inproceedings{CannaviccioABM_WebDB2018, author = {Matteo Cannaviccio and Lorenzo Ariemma and Denilson Barbosa and Paolo Merialdo}, title = {Leveraging Wikipedia Table Schemas for Knowledge Graph Augmentation}, booktitle = {Proceedings of the 21st International Workshop on Web and Databases, Houston, TX, USA, June 10th, 2018}, year = {2018}, doi = {10.1145/3201463.3201468}, pages = {5:1-5:6}, publisher = {{ACM}} } - Matching Resumes to Job Descriptions with Stacked ModelsIn Advances in Artificial Intelligence - 31st Canadian Conference on Artificial Intelligence, Canadian AI 2018, Toronto, ON, Canada, May 8-11, 2018, Proceedings, pp. 304–309, 2018.
@inproceedings{XuBarbosa_CanAI_2018, author = {Peng Xu and Denilson Barbosa}, title = {Matching Resumes to Job Descriptions with Stacked Models}, booktitle = {Advances in Artificial Intelligence - 31st Canadian Conference on Artificial Intelligence, Canadian {AI} 2018, Toronto, ON, Canada, May 8-11, 2018, Proceedings}, year = {2018}, doi = {10.1007/978-3-319-89656-4_31}, pages = {304--309} } - Towards Annotating Relational Data on the Web with Language ModelsIn Proceedings of The Web Conference 2018, pp. 1307–1316, 2018.
Tables and structured lists on Web pages are a potential source of valuable information, and several methods have been proposed to annotate them with semantics that can be leveraged for search, question answering and information extraction. This paper is concerned with the specific problem of finding and ranking relations from a given Knowledge Graph (KG) that hold over pairs of entities juxtaposed in a table or structured list. The state-of-the-art for this task is to attempt to link the entities mentioned in the table cells to objects in the KG and rank the relations that hold for those linked objects. As a result, these methods are hampered by the incompleteness and uneven coverage in even the best knowledge graphs available today. The alternative described here does not require entity linking, relying instead on ranking relations using generative language models derived from Web-scale corpora. As such, it can produce quality results even when the entities in the table are missing in the KG. The experimental validation, designed to expose the challenges posed by KG incompleteness, shows that our approach is robust and effective in practice.
@inproceedings{Cannaviccio_et_al_2018, author = {Matteo Cannaviccio and Denilson Barbosa and Paolo Merialdo}, title = {Towards Annotating Relational Data on the Web with Language Models}, booktitle = {Proceedings of The Web Conference 2018}, year = {2018}, doi = {10.1145/3178876.3186029}, pages = {1307--1316} } - Robust named entity disambiguation with random walksSemantic Web, 9(4), pp. 459–479, 2018.
Named Entity Disambiguation is the task of assigning entities from a Knowledge Graph (KG) to mentions of such entities in a textual document. The state-of-the-art for this task balances two disparate sources of similarity: lexical, defined as the pairwise similarity between mentions in the text and names of entities in the KG; and semantic, defined through some graph-theoretic property of a subgraph of the KG induced by the choice of entities for each mention. Departing from previous work, our notion of semantic similarity is rooted in Information Theory and is defined as the mutual information between random walks on the disambiguation graph induced by choice of entities for each mention. We describe an iterative algorithm based on this idea, and show an extension that uses learning-to-rank, which yields further improvements. Our experimental evaluation demonstrates that this approach is robust and very competitive on well-known existing benchmarks. We also justify the need for new and more difficult benchmarks, and provide an extensive experimental comparison of our method and previous work on these new benchmarks.
@article{Guo_Barbosa_2018, author = {Zhaochen Guo and Denilson Barbosa}, title = {Robust named entity disambiguation with random walks}, journal = {Semantic Web}, year = {2018}, doi = {10.3233/SW-170273}, number = {4}, pages = {459--479}, volume = {9} }
2016
- Relationship Queries on Extended Knowledge GraphsIn Proceedings of the Ninth ACM International Conference on Web Search and Data Mining, San Francisco, CA, USA, February 22-25, 2016, pp. 605–614, 2016.
Entity search over text corpora is not geared for relationship queries where answers are tuples of related entities and where a query often requires joining cues from multiple documents. With large knowledge graphs, structured querying on their relational facts is an alternative, but often suffers from poor recall because of mismatches between user queries and the knowledge graph or because of weakly populated relations.
@inproceedings{Yahya_etal_WSDM2016, author = {Mohamed Yahya and Denilson Barbosa and Klaus Berberich and Qiuyue Wang and Gerhard Weikum}, title = {Relationship Queries on Extended Knowledge Graphs}, booktitle = {Proceedings of the Ninth {ACM} International Conference on Web Search and Data Mining, San Francisco, CA, USA, February 22-25, 2016}, year = {2016}, doi = {10.1145/2835776.2835795}, pages = {605--614}, publisher = {{ACM}} } - Hadoop energy consumption reduction with hybrid HDFSIn Proceedings of the 31st Annual ACM Symposium on Applied Computing, Pisa, Italy, April 4-8, 2016, pp. 406–411, 2016.
Apache Hadoop has evolved significantly over the last years, with more than 60 releases bringing new features. By implementing the MapReduce programming paradigm and leveraging HDFS, its distributed file system, Hadoop has become a reliable and fault tolerant middleware for parallel and distributed computing over large datasets. Nevertheless, Hadoop may struggle under certain workloads, resulting in poor performance and high energy consumption. Users increasingly demand that high performance computing solutions address sustainability and limit energy consumption. In this paper, we introduce HDFSH, a hybrid storage mechanism for HDFS, which uses a combination of Hard Disks and Solid-State Disks to achieve higher performance while saving power in Hadoop computations. HDFSH brings to the middleware the best from HDs (affordable cost per GB and high storage capacity) and SSDs (high throughput and low energy consumption) in a configurable fashion, using dedicated storage zones for each storage device type. We implemented our mechanism as a block placement policy for HDFS, and assessed it over six recent releases of Hadoop with different architectural properties. Results indicate that our approach increases overall job performance while decreasing the energy consumption under most hybrid configurations evaluated. Our results also showed that, in many cases, storing only part of the data in SSDs results in significant energy savings and execution speedups.
@inproceedings{PolatoBHK_SAC2016, author = {Ivanilton Polato and Denilson Barbosa and Abram Hindle and Fabio Kon}, title = {Hadoop energy consumption reduction with hybrid {HDFS}}, booktitle = {Proceedings of the 31st Annual {ACM} Symposium on Applied Computing, Pisa, Italy, April 4-8, 2016}, year = {2016}, doi = {10.1145/2851613.2851623}, editor = {Sascha Ossowski}, pages = {406--411}, publisher = {{ACM}} } - Accurate fact harvesting from natural language text in wikipedia with LectorIn Proceedings of the 19th International Workshop on Web and Databases, San Francisco, CA, USA, June 26, 2016, pp. 9, 2016.
Many approaches have been introduced recently to automatically create or augment Knowledge Graphs (KGs) with facts extracted from Wikipedia, particularly its structured components like the infoboxes. Although these structures are valuable, they represent only a fraction of the actual information expressed in the articles. In this work, we quantify the number of highly accurate facts that can be harvested with high precision from the text of Wikipedia articles using information extraction techniques bootstrapped from the entities and relations already in a KG. Our experimental evaluation, which uses Freebase as reference KG, reveals we can augment several relations in the domain of people by more than 10%, with facts whose accuracy are over 95%. Moreover, the vast majority of these facts are missing from the infoboxes, YAGO and DBpedia.
@inproceedings{CannaviccioBM_WebDB2016, author = {Matteo Cannaviccio and Denilson Barbosa and Paolo Merialdo}, title = {Accurate fact harvesting from natural language text in wikipedia with Lector}, booktitle = {Proceedings of the 19th International Workshop on Web and Databases, San Francisco, CA, USA, June 26, 2016}, year = {2016}, doi = {10.1145/2932194.2932203}, pages = {9}, publisher = {{ACM}} }
2015
- Hadoop branching: Architectural impacts on energy and performanceIn Sixth International Green and Sustainable Computing Conference, IGSC 2015, Las Vegas, NV, USA, December 14-16, 2015, pp. 1–4, 2015.
Data centers are notorious energy consumers. In fact, studies have shown that for every $1 spent on hardware in the datacenter, $0.50 is spent on powering this hardware over its lifetime. Data centers host real or virtual (i.e., cloud) clusters that often execute large compute jobs using MapReduce, of which Hadoop is a popular implementation. Like other successful open source projects, Hadoop has been maintained and evolved over time with new resource management features being added over time in an effort to improve performance, raising questions as to whether such architectural evolution has achieved its goal, and if so, at what cost. In this work we apply Green Mining to find out that later versions of Hadoop - who exhibit more dynamic resource control - can suffer from serious energy consumption performance regressions.
@inproceedings{PolatoBHK_Green2015, author = {Ivanilton Polato and Denilson Barbosa and Abram Hindle and Fabio Kon}, title = {Hadoop branching: Architectural impacts on energy and performance}, booktitle = {Sixth International Green and Sustainable Computing Conference, {IGSC} 2015, Las Vegas, NV, USA, December 14-16, 2015}, year = {2015}, doi = {10.1109/IGCC.2015.7393709}, pages = {1--4}, publisher = {{IEEE} Computer Society} } - Identifying Controversial Wikipedia Articles Using Editor Collaboration NetworksACM TIST, 6(1), pp. 5, 2015.
Wikipedia is probably the most commonly used knowledge reference nowadays, and the high quality of its articles is widely acknowledged. Nevertheless, disagreement among editors often causes some articles to become controversial over time. These articles span thousands of popular topics, including religion, history, and politics, to name a few, and are manually tagged as controversial by the editors, which is clearly suboptimal. Moreover, disagreement, bias, and conflict are expressed quite differently in Wikipedia compared to other social media, rendering previous approaches ineffective. On the other hand, the social process of editing Wikipedia is partially captured in the edit history of the articles, opening the door for novel approaches. This article describes a novel controversy model that builds on the interaction history of the editors and not only predicts controversy but also sheds light on the process that leads to controversy. The model considers the collaboration history of pairs of editors to predict their attitude toward one another. This is done in a supervised way, where the votes of Wikipedia administrator elections are used as labels indicating agreement (i.e., support vote) or disagreement (i.e., oppose vote). From each article, a collaboration network is built, capturing the pairwise attitude among editors, allowing the accurate detection of controversy. Extensive experimental results establish the superiority of this approach compared to previous work and very competitive baselines on a wide range of settings.
@article{Rad_Barbosa_TIST2015, author = {Hoda Sepehri-Rad and Denilson Barbosa}, title = {Identifying Controversial Wikipedia Articles Using Editor Collaboration Networks}, journal = {ACM TIST}, year = {2015}, doi = {10.1145/2630075}, number = {1}, pages = {5}, volume = {6} } - Inferencing in Information Extraction: Techniques and ApplicationsIn IEEE 31st International Conference on Data Engineering, 2015.
Information extraction at Web scale has become one of the most important research topics in data management since major commercial search engines started incorporating knowledge in their search results a couple of years ago [1]. Users increasingly expect structured knowledge as answers to their search needs. Using Bing as an example, the result page for “Lionel Messi” is full of structured knowledge facts, such as his birthday and awards. The research efforts towards improving the accuracy and coverage of such knowledge bases have led to significant advances in Information Extraction techniques [2], [3]. As the initial challenge of accurately extracting facts for popular entities are being addressed, more difficult challenges have emerged such as extending knowledge coverage to long tail entities and domains, understanding interestingness and usefulness of facts within a given context, and addressing information-seeking needs more directly and accurately. In this tutorial, we will survey the recent research efforts and provide an introduction to the techniques that address those challenges, and the applications that benefit from the adoption of those techniques. In particular, this tutorial will focus on a variety of techniques that can be broadly viewed as knowledge inferencing, i.e., combining multiple data sources and extraction techniques to verify existing knowledge and derive new knowledge. More specifically, we focus on four main categories of inferencing techniques: 1) deep natural language processing using machine learning techniques, 2) data cleaning using integrity constraints, 3) large-scale probabilistic reasoning, and 4) leveraging human expertise for domain knowledge extraction.
@inproceedings{Barbosa_etal_ICDE2015, author = {Denilson Barbosa and Haixun Wang and Cong Yu}, title = {Inferencing in Information Extraction: Techniques and Applications}, booktitle = {IEEE 31st International Conference on Data Engineering}, year = {2015}, doi = {10.1109/icde.2015.7113420}, publisher = {IEEE} } - Advances in Artificial Intelligence - 28th Canadian Conference on Artificial Intelligence, Canadian AI 2015, Halifax, Nova Scotia, Canada, June 2-5, 2015, ProceedingsLecture Notes in Computer Science, Springer, 2015.
@proceedings{Proc_Canadian_AI_2015, title = {Advances in Artificial Intelligence - 28th Canadian Conference on Artificial Intelligence, Canadian {AI} 2015, Halifax, Nova Scotia, Canada, June 2-5, 2015, Proceedings}, year = {2015}, doi = {10.1007/978-3-319-18356-5}, editor = {Denilson Barbosa and Evangelos E. Milios}, isbn = {978-3-319-18355-8}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, volume = {9091} } - Entity Recognition and Linking on Tweets with Random WalksIn Proceedings of the the 5th Workshop on Making Sense of Microposts co-located with the 24th International World Wide Web Conference, Florence, Italy, May 18th, 2015., pp. 57–58, 2015.
This paper presents our system at the #Microposts2015 NEEL Challenge [4]. The task is to recognize and type mentions from English Microposts, and link them to their corresponding entries in DBpedia 2014. For this task, we developed a method based on a state-of-the-art entity linking system - REL-RW [2], which exploits the entity graph from the knowledge base to compute semantic relatedness between entities, and use it for entity disambiguation. The advantage of the approach is its robustness for various types of documents. We built our system on REL-RW and employed a tweet specific NER component to improve the performance on tweets. The system achieved overall 0.35 F1 on the development dataset from NEEL 2015, while the disambiguation component alone can achieve 0.70 F1.
@inproceedings{Guo_Barbosa_MSM2015, author = {Zhaochen Guo and Denilson Barbosa}, title = {Entity Recognition and Linking on Tweets with Random Walks}, booktitle = {Proceedings of the the 5th Workshop on Making Sense of Microposts co-located with the 24th International World Wide Web Conference, Florence, Italy, May 18th, 2015.}, year = {2015}, pages = {57--58} } - Hybrid HDFS: decreasing energy consumption and speeding up Hadoop using SSDsPeerJ PrePrints, 3, pp. e1320, 2015.
Apache Hadoop has evolved significantly over the last years, with more than 60 releases bringing new features. By implementing the MapReduce programming paradigm and leveraging HDFS, its distributed file system, Hadoop has become a reliable and fault tolerant middleware for parallel and distributed computing over large datasets. Nevertheless, Hadoop may struggle under certain workloads, resulting in poor performance and high energy consumption. Users increasingly demand that high performance computing solutions being to address sustainability and limit power consumption. In this paper, we introduce HDFSH , a hybrid storage mechanism for HDFS, which uses a combination of Hard Disks and Solid-State Disks to achieve higher performance while saving power in Hadoop computations. HDFSH brings to middleware the best from HDs (affordable cost per GB and high storage capacity) and SSDs (high throughput and low energy consumption) in a configurable fashion, using dedicated storage zones for each storage device type. We implemented our mechanism as a block placement policy for HDFS, and assessed it over six recent releases of the Hadoop project, representing different designs of the Hadoop middleware. Results indicate that our approach increases overall job performance while decreasing the energy consumption under most hybrid configurations evaluated. Our results also showed that in many cases storing only part of the data in SSDs results in significant energy savings and execution speedups.
@article{PolatoBHK_PeerJ2015, author = {Ivanilton Polato and Denilson Barbosa and Abram Hindle and Fabio Kon}, title = {Hybrid {HDFS:} decreasing energy consumption and speeding up Hadoop using SSDs}, journal = {PeerJ PrePrints}, year = {2015}, doi = {10.7287/peerj.preprints.1320v1}, pages = {e1320}, volume = {3} }
2014
- Robust Entity Linking via Random WalksIn Proceedings of the 23rd ACM International Conference on Conference on Information and Knowledge Management, CIKM 2014, Shanghai, China, November 3-7, 2014, pp. 499–508, 2014.
Entity Linking is the task of assigning entities from a Knowledge Base to textual mentions of such entities in a document. State-of-the-art approaches rely on lexical and statistical features which are abundant for popular entities but sparse for unpopular ones, resulting in a clear bias towards popular entities and poor accuracy for less popular ones. In this work, we present a novel approach that is guided by a natural notion of semantic similarity which is less amenable to such bias. We adopt a unified semantic representation for entities and documents - the probability distribution obtained from a random walk on a subgraph of the knowledge base - which can overcome the feature sparsity issue that affects previous work. Our algorithm continuously updates the semantic signature of the document as mentions are disambiguated, thus focusing the search based on context. Our experimental evaluation uses well-known benchmarks and different samples of a Wikipedia-based benchmark with varying entity popularity; the results illustrate well the bias of previous methods and the superiority of our approach, especially for the less popular entities.
@inproceedings{Guo_Barbosa_CIKM2014, author = {Zhaochen Guo and Denilson Barbosa}, title = {Robust Entity Linking via Random Walks}, booktitle = {Proceedings of the 23rd {ACM} International Conference on Conference on Information and Knowledge Management, {CIKM} 2014, Shanghai, China, November 3-7, 2014}, year = {2014}, doi = {10.1145/2661829.2661887}, pages = {499--508}, publisher = {{ACM}} } - Entity linking with a unified semantic representationIn 23rd International World Wide Web Conference, WWW '14, Seoul, Republic of Korea, April 7-11, 2014, Companion Volume, pp. 1305-1310, 2014.
Entity Linking (EL) consists in linking mentions in a document to their referent entities in a Knowledge Base. Current approaches fall into two main categories: local approaches, in which mentions are linked independently of each other, and global approaches, in which all mentions are linked collectively. Local approaches often ignore the semantic relatedness of entities, and while global approaches incorporate the semantic relatedness, they tend to focus only on directly connected entities, ignoring indirect connections which might be useful. We present a global EL approach that unifies the representation of the semantics of entities and documents--the probability distribution of entities being visited during a random walk on an entity graph--that accounts for direct and indirect connections. An experimental evaluation shows that our method outperforms five state-of-the-art EL systems and two very strong baselines.
@inproceedings{Guo_Barbosa_WWW2014, author = {Zhaochen Guo and Denilson Barbosa}, title = {Entity linking with a unified semantic representation}, booktitle = {23rd International World Wide Web Conference, WWW '14, Seoul, Republic of Korea, April 7-11, 2014, Companion Volume}, year = {2014}, doi = {10.1145/2567948.2579705}, pages = {1305-1310}, publisher = {ACM} } - Knowledge Base Augmentation using Tabular DataIn 7th Workshop on Linked Data on the Web, 2014.
Large linked data repositories have been built by leveraging semi-structured data in Wikipedia (e.g., DBpedia) and through extracting information from natural language text (e.g., YAGO). However, the Web contains many other vast sources of linked data, such as structured HTML tables and spreadsheets. Often, the semantics in such tables is hidden, preventing one from extracting triples from them directly. This paper describes a probabilistic method that augments an existing knowledge base with facts from tabular data by leveraging a Web text corpus and natural language patterns associated with relations in the knowledge base. A prelim-inary evaluation shows high potential for this technique in augmenting linked data repositories. 1.
@inproceedings{Sekhavat_etal_LDOW2014, author = {Yoones A. Sekhavat and Francesco di Paolo and Denilson Barbosa and Paolo Merialdo}, title = {Knowledge Base Augmentation using Tabular Data}, booktitle = {7th Workshop on Linked Data on the Web}, year = {2014}, publisher = {CEUR-WS.org} } - Improving Open Relation Extraction via Sentence Re-StructuringIn Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14), 2014.
Information Extraction is an important task in Natural Language Processing, consisting of finding a structured representation for the information expressed in natural language text. Two key steps in information extraction are identifying the entities mentioned in the text, and the relations among those entities. In the context of Information Extraction for the World Wide Web, unsupervised relation extraction methods, also called Open Relation Extraction (ORE) systems, have become prevalent, due to their effectiveness without domain-specific training data. In general, these systems exploit part-of-speech tags or semantic information from the sentences to determine whether or not a relation exists, and if so, its predicate. This paper discusses some of the issues that arise when even moderately complex sentences are fed into ORE systems. A process for re-structuring such sentences is discussed and evaluated. The proposed approach replaces complex sentences by several others that, together, convey the same meaning and are more amenable to extraction by current ORE systems. The results of an experimental evaluation show that this approach succeeds in reducing the processing time and increasing the accuracy of the state-of-the-art ORE systems.
@inproceedings{Schmidek_Barbosa_LREC2014, author = {Jordan Schmidek and Denilson Barbosa}, title = {Improving Open Relation Extraction via Sentence Re-Structuring}, booktitle = {Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14)}, year = {2014}, address = {Reykjavik, Iceland}, editor = {Nicoletta Calzolari (Conference Chair) and Khalid Choukri and Thierry Declerck and Hrafn Loftsson and Bente Maegaard and Joseph Mariani and Asuncion Moreno and Jan Odijk and Stelios Piperidis}, isbn = {978-2-9517408-8-4}, publisher = {European Language Resources Association (ELRA)}, url = {https://aclanthology.org/L14-1029/} } - LDC2014T27 - Benchmarks for Open Relation ExtractionLinguistic Data Consortium, 2014.
@misc{Mesquita_et_al_LDC2014, author = {Filipe Mesquita and Jordan Schmidek and Denilson Barbosa}, title = {LDC2014T27 - Benchmarks for Open Relation Extraction}, year = {2014}, date-modified = {2015-10-01 11:52:57 +0000}, month = {December}, published = {Linguistic Data Consortium} } - Extracting Family Relationship Networks from NovelsCoRR, abs/1405.0603, 2014.
We present an approach to the extraction of family relations from literary narrative, which incorporates a technique for utterance attribution proposed recently by Elson and McKeown (2010). In our work this technique is used in combination with the detection of vocatives - the explicit forms of address used by the characters in a novel. We take advantage of the fact that certain vocatives indicate family relations between speakers. The extracted relations are then propagated using a set of rules. We report the results of the application of our method to Jane Austen's Pride and Prejudice.
@article{Makazhanov_etal_CORR2014, author = {Aibek Makazhanov and Denilson Barbosa and Grzegorz Kondrak}, title = {Extracting Family Relationship Networks from Novels}, journal = {CoRR}, year = {2014}, doi = {10.48550/arxiv.1405.0603}, eprint = {1405.0603}, eprinttype = {arXiv}, volume = {abs/1405.0603} } - Sentiment Analysis for Streams of Web Data: A Case Study of Brazilian Financial MarketsIn Proceedings of the 20th Brazilian Symposium on Multimedia and the Web, WebMedia 2014, João Pessoa, Brazil, November 18-21, 2014, pp. 167–170, 2014.
With the rise of Web 2.0 applications, most people started consuming information and sharing opinions and ideas about most aspects of their lives on a variety of social media platforms, creating massive and continuous streams of valuable data. While this opened the door for information extraction and mining techniques that can help us understand different aspects of society, extracting useful information from such streams of Web data is far from trivial. In this setting, sentiment analysis techniques can be convenient as they are capable of summarizing general feeling about entities people care about, such as products and companies. Therefore, they can be quite applicable in scenarios like the stock market, which also has tremendous impact on society. This paper describes and evaluates two different techniques for sentiment analysis applied to the Brazilian stock market data: lexicon-based and machine learning based, considering a wide range of text pre-processing and feature selection approaches.
@inproceedings{Neuenschwander_etal_WebMedia2014, author = {Bruna Neuenschwander and Adriano C{\'{e}}sar Machado Pereira and Wagner Meira Jr. and Denilson Barbosa}, title = {Sentiment Analysis for Streams of Web Data: {A} Case Study of Brazilian Financial Markets}, booktitle = {Proceedings of the 20th Brazilian Symposium on Multimedia and the Web, WebMedia 2014, Jo{\~{a}}o Pessoa, Brazil, November 18-21, 2014}, year = {2014}, doi = {10.1145/2664551.2664579}, pages = {167--170}, publisher = {ACM} } - Using triads to identify local community structure in social networksIn 2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2014, Beijing, China, August 17-20, 2014, pp. 108–112, 2014.
We present our novel community mining algorithm that uses only local information to accurately identify communities, outliers, and hubs in social networks. The main component of our algorithm is the T metric, which evaluates the relative quality of a community by considering the number of internal and external triads (3-node cliques) it contains. Furthermore we propose an intuitive statistical method based on our T metric, which correctly identifies outlier and hub nodes within each discovered community. Finally, we evaluate our approach on a series of ground-truth networks and show that our method outperforms the state-of-the-art in community mining algorithms.
@inproceedings{Fagnan_et_al_ASONAM14, author = {Justin Fagnan and Osmar R. Zaiane and Denilson Barbosa}, title = {Using triads to identify local community structure in social networks}, booktitle = {2014 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining, ASONAM 2014, Beijing, China, August 17-20, 2014}, year = {2014}, doi = {10.1109/ASONAM.2014.6921568}, pages = {108--112} }
2013
- Automatically Training Form ClassifiersIn Proceedings of the 14th International Web Information Systems Engineering Conference, pp. 441-453, 2013.
@inproceedings{Moraes_etal_WISE2013, author = {Mauricio C. Moraes and Carlos A. Heuser and Viviane Pereira Moreira and Denilson Barbosa}, title = {Automatically Training Form Classifiers}, booktitle = {Proceedings of the 14th International Web Information Systems Engineering Conference}, year = {2013}, doi = {10.1007/978-3-642-41230-1_37}, pages = {441-453} } - Effectiveness and Efficiency of Open Relation ExtractionIn Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pp. 447–457, 2013.
A large number of Open Relation Extraction approaches have been proposed recently, covering a wide range of NLP machinery, from “shallow” (e.g., part-of-speech tagging) to “deep” (e.g., semantic role labeling‐SRL). A natural question then is what is the tradeoff between NLP depth (and associated computational cost) versus effectiveness. This paper presents a fair and objective experimental comparison of 8 state-of-the-art approaches over 5 different datasets, and sheds some light on the issue. The paper also describes a novel method, EXEMPLAR, which adapts ideas from SRL to less costly NLP machinery, resulting in substantial gains both in efficiency and effectiveness, over binary and n-ary relation extraction tasks.
@inproceedings{Mesquita_etal_EMNLP2013, author = {Filipe Mesquita and Jordan Schmidek and Denilson Barbosa}, title = {Effectiveness and Efficiency of Open Relation Extraction}, booktitle = {Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing}, year = {2013}, doi = {10.18653/v1/d13-1043}, month = {October}, pages = {447--457}, publisher = {Association for Computational Linguistics} } - Identification of Speakers in NovelsIn Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 1312–1320, 2013.
Speaker identification is the task of attributing utterances to characters in a literary narrative. It is challenging to automate because the speakers of the majority of utterances are not explicitly identified in novels. In this paper, we present a supervised machine learning approach for the task that incorporates several novel features. The experimental results show that our method is more accurate and general than previous approaches to the problem. 1
@inproceedings{He_et_al_ACL2013, author = {Hua He and Denilson Barbosa and Grzegorz Kondrak}, title = {Identification of Speakers in Novels}, booktitle = {Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)}, year = {2013}, address = {Sofia, Bulgaria}, month = {August}, pages = {1312--1320}, publisher = {Association for Computational Linguistics}, url = {https://aclanthology.org/P13-1129/} } - Open Information Extraction with Tree KernelsIn Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pp. 868–877, 2013.
Traditional relation extraction seeks to identify pre-specified semantic relations within natural language text, while open Information Extraction (Open IE) takes a more general approach, and looks for a variety of relations without restriction to a fixed relation set. With this generalization comes the question, what is a relation? For example, should the more general task be restricted to relations mediated by verbs, nouns, or both? To help answer this question, we propose two levels of subtasks for Open IE. One task is to determine if a sentence potentially contains a relation between two entities? The other task looks to confirm explicit relation words for two entities. We propose multiple SVM models with dependency tree kernels for both tasks. For explicit relation extraction, our system can extract both noun and verb relations. Our results on three datasets show that our system is superior when compared to state-of-the-art systems like REVERB and OLLIE for both tasks. For example, in some experiments our system achieves 33 % improvement on nominal relation extraction over OLLIE. In addition we propose an unsupervised rule-based approach which can serve as a strong baseline for Open IE systems. 1
@inproceedings{Xu_et_al_NAACL_HLT2_2013, author = {Ying Xu and Mi-Young Kim and Kevin Quinn and Randy Goebel and Denilson Barbosa}, title = {Open Information Extraction with Tree Kernels}, booktitle = {Proceedings of the 2013 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies}, year = {2013}, address = {Atlanta, Georgia}, month = {June}, pages = {868--877}, publisher = {Association for Computational Linguistics}, url = {https://aclanthology.org/N13-1107/} } - Perspectives on Business IntelligenceSynthesis Lectures on Data Management, Morgan and Claypool Publishers, 2013.
@book{BIN_book_2013, author = {Raymond T. Ng and Patricia C. Arocena and Denilson Barbosa and Giuseppe Carenini and Luiz Gomes Jr. and Stephan Jou and Rock Anthony Leung and Evangelos Milios and Renee J. Miller and John Mylopoulos and Rachel A. Pottinger and Frank Tompa and Eric Yu}, title = {Perspectives on Business Intelligence}, year = {2013}, doi = {10.2200/s00491ed1v01y201303dtm034}, publisher = {Morgan and Claypool Publishers}, series = {Synthesis Lectures on Data Management} } - Shallow Information Extraction for the Knowledge WebIn IEEE 29th International Conference on Data Engineering, pp. 1264 - 1267, 2013.
A new breed of Information Extraction tools has become popular and shown to be very effective in building massive-scale knowledge bases that fuel applications such as question answering and semantic search. These approaches rely on Web-scale probabilistic models populated through shallow language processing of the text, pre-existing knowledge, and structured data already on the Web. This tutorial provides an introduction to these techniques, starting from the foundations of information extraction, and covering some of its key applications.
@inproceedings{Barbosa_etal_ICDE2013, author = {Denilson Barbosa and Haixun Wang and Cong Yu}, title = {Shallow Information Extraction for the Knowledge Web}, booktitle = {IEEE 29th International Conference on Data Engineering}, year = {2013}, doi = {10.1109/icde.2013.6544920}, pages = {1264 - 1267}, publisher = {IEEE} } - SociQL: A Query Language for the SocialWebIn Advances in Network Analysis and its Applications, Springer Berlin Heidelberg, pp. 381-406, 2013.
@incollection{Serrano_etal_ANAA2013, author = {Diego Serrano and Eleni Stroulia and Denilson Barbosa and Victor Guana}, title = {{SociQL: A Query Language for the SocialWeb}}, booktitle = {Advances in Network Analysis and its Applications}, year = {2013}, doi = {10.1007/978-3-642-30904-5_17}, editor = {Kranakis, Evangelos}, pages = {381-406}, publisher = {Springer Berlin Heidelberg}, series = {Mathematics in Industry}, volume = {18} } - Pre-Query Discovery of Domain-specific Query Forms: A SurveyIEEE Transactions on Knowledge and Data Engineering, 25(8), pp. 1830-1848, 2013.
The discovery of HTML query forms is one of the main challenges in Deep Web crawling. Automatic solutions for this problem perform two main tasks. The first is locating HTML forms on the Web, which is done through the use of traditional/focused crawlers. The second is identifying which of these forms are indeed meant for querying, which also typically involves determining a domain for the underlying data source (and thus for the form as well). This problem has attracted a great deal of interest, resulting in a long list of algorithms and techniques. Some methods submit requests through the forms and then analyze the data retrieved in response, typically requiring a great deal of knowledge about the domain as well as semantic processing. Others do not employ form submission, to avoid such difficulties, although some techniques rely to some extent on semantics and domain knowledge. This survey gives an up-to-date review of methods for the discovery of domain-specific query forms that do not involve form submission. We detail these methods and discuss how form discovery has become increasingly more automated over time. We conclude with a forecast of what we believe are the immediate next steps in this trend.
@article{Moraes_etal_TKDE2013, author = {Mauricio Moraes and Carlos A. Heuser and Viviane Moreira and Denilson Barbosa}, title = {{Pre-Query Discovery of Domain-specific Query Forms: A Survey}}, journal = {IEEE Transactions on Knowledge and Data Engineering}, year = {2013}, doi = {10.1109/TKDE.2012.111}, number = {8}, pages = {1830-1848}, volume = {25} }
2012
- Identifying controversial articles in Wikipedia: A comparative studyIn 8th International Symposium on Wikis and Open Collaboration, pp. 7:1–7:10, 2012.
Wikipedia articles are the result of the collaborative editing of a diverse group of anonymous volunteer editors, who are passionate and knowledgeable about specific topics. One can argue that this plurality of perspectives leads to broader coverage of the topic, thus benefitting the reader. On the other hand, differences among editors on polarizing topics can lead to controversial or questionable content, where facts and arguments are presented and discussed to support a particular point of view. Controversial articles are manually tagged by Wikipedia editors, and span many interesting and popular topics, such as religion, history, and politics, to name a few. Recent works have been proposed on automatically identifying controversy within unmarked articles. However, to date, no systematic comparison of these efforts has been made. This is in part because the various methods are evaluated using different criteria and on different sets of articles by different authors, making it hard for anyone to verify the efficacy and compare all alternatives. We provide a first attempt at bridging this gap. We compare five different methods for modelling and identifying controversy, and discuss some of the unique difficulties and opportunities inherent to the way Wikipedia is produced.
@inproceedings{Rab_Barbosa_Wikisym2012, author = {Hoda Sepehri-Rad and Denilson Barbosa}, title = {Identifying controversial articles in Wikipedia: A comparative study}, booktitle = {8th International Symposium on Wikis and Open Collaboration}, year = {2012}, doi = {10.1145/2462932.2462942}, pages = {7:1--7:10}, publisher = {ACM} } - Towards scalable summarization and visualization of large text corpora (abstract only)In Proceedings of the ACM SIGMOD International Conference on Management of Data, pp. 863, 2012.
Society is awash with problems requiring the analysis of vast quantities of text and data. From detecting flu trends out of twitter conversations to finding scholarly works answering specific questions, we rely more and more on computers to process text for us. Text analytics is the application of computational, mathematical, and statistical models to derive information from large quantities of data coming primarily as text. Our project provides fast and effective text-analytics tools for large document collections, such as the blogosphere. We use natural language processing and database techniques to extract, collect, analyze, visualize, and archive information extracted from text. We focus on discovering relationships between entities (people, places, organizations, etc.) mentioned in one or more sources (blog posts or news articles). We built a custom solution using mostly off-the-shelf, open-source tools to provide a scalable platform for users to search and analyze large text corpora. Currently, we provide two main outlets for users to discover these relations: (1) full-text search over the documents and (2) graph visualizations of the entities and their relationships. This provides the user with succinct and easily digestible information gleaned from the corpus as a whole. For example, we can easily pose queries like which companies were bought by Google? as entity:google relation:bought. The extracted data is stored on a combination of the noSQL database CouchDB and Apache's Lucene. This combination is justified as our work-flow consists of offline batch insertions with almost no updates. Because we support specialized queries, we can forgo the flexibility of traditional SQL solutions and materialize all necessary indices, which are used to quickly query large amounts of de-normalized data using MapReduce. Lucene provides a flexible and powerful query syntax to yield relevant ranked results to the user. Moreover, its indices are synchronized by a process subscribed to the list of database changes published by CouchDB. The graph visualizations rely on CouchDB's ability to export the data in any format: we currently use a customized graph visualization relying on XML data. Finally, we use memcached to further improve the performance, especially for queries involving popular entities.
@inproceedings{Sliwkanich_etal_SIGMOD2012, author = {Tyler Sliwkanich and Douglas Schneider and Aaron Yong and Mitchell Home and Denilson Barbosa}, title = {Towards scalable summarization and visualization of large text corpora (abstract only)}, booktitle = {Proceedings of the ACM SIGMOD International Conference on Management of Data}, year = {2012}, crossref = {conf/sigmod/2012}, doi = {10.1145/2213836.2213970}, ee = {http://doi.acm.org/10.1145/2213836.2213970}, pages = {863} } - Automatic Evaluation of Relation Extraction Systems on Large-scaleIn Proceedings of the Joint Workshop on Automatic Knowledge Base Construction and Web-scale Knowledge Extraction (AKBC-WEKEX), pp. 19–24, 2012.
The extraction of relations between named entities from natural language text is a longstanding challenge in information extraction, especially in large-scale. A major challenge for the advancement of this research field has been the lack of meaningful evaluation frameworks based on realistic-sized corpora. In this paper we propose a framework for large-scale evaluation of relation extraction systems based on an automatic annotator that uses a public online database and a large web corpus. 1
@inproceedings{Bronzi_etal_AKBCWEKEX2012, author = {Mirko Bronzi and Zhaochen Guo and Filipe Mesquita and Denilson Barbosa and Paolo Merialdo}, title = {Automatic Evaluation of Relation Extraction Systems on Large-scale}, booktitle = {Proceedings of the Joint Workshop on Automatic Knowledge Base Construction and Web-scale Knowledge Extraction (AKBC-WEKEX)}, year = {2012}, address = {Montreal, Canada}, month = {June}, pages = {19--24}, publisher = {Association for Computational Linguistics}, url = {https://aclanthology.org/W12-3004/} } - Extracting information networks from the blogosphereACM Transactions on the Web, 6(3), pp. 11:1–11:33, 2012.
We study the problem of automatically extracting information networks formed by recognizable entities as well as relations among them from social media sites. Our approach consists of using state-of-the-art natural language processing tools to identify entities and extract sentences that relate such entities, followed by using text-clustering algorithms to identify the relations within the information network. We propose a new term-weighting scheme that significantly improves on the state-of-the-art in the task of relation extraction, both when used in conjunction with the standard tf ċ idf scheme and also when used as a pruning filter. We describe an effective method for identifying benchmarks for open information extraction that relies on a curated online database that is comparable to the hand-crafted evaluation datasets in the literature. From this benchmark, we derive a much larger dataset which mimics realistic conditions for the task of open information extraction. We report on extensive experiments on both datasets, which not only shed light on the accuracy levels achieved by state-of-the-art open information extraction tools, but also on how to tune such tools for better results.
@article{Merhav_etal_TWEB2012, author = {Yuval Merhav and Filipe Mesquita and Denilson Barbosa and Wai Gen Yee and Ophir Frieder}, title = {Extracting information networks from the blogosphere}, journal = {ACM Transactions on the Web}, year = {2012}, acmid = {2344418}, address = {New York, NY, USA}, articleno = {11}, doi = {10.1145/2344416.2344418}, issn = {1559-1131}, issue_date = {September 2012}, keywords = {clustering, domain frequency, named entities, open information extraction, relation extraction}, number = {3}, numpages = {33}, pages = {11:1--11:33}, publisher = {ACM}, volume = {6} } - Leveraging editor collaboration patterns in wikipediaIn Proceedings of the 23rd ACM conference on Hypertext and social media, pp. 13–22, 2012.
Predicting the positive or negative attitude of individuals towards each other in a social environment has long been of interest, with applications in many domains. We investigate this problem in the context of the collaborative editing of articles in Wikipedia, showing that there is enough information in the edit history of the articles that can be utilized for predicting the attitude of co-editors. We train a model using a distant supervision approach, by labeling interactions between editors as positive or negative depending on how these editors vote for each other in Wikipedia admin elections. We use the model to predict the attitude among other editors, who have neither run nor voted in an election. We validate our model by assessing its accuracy in the tasks of predicting the results of the actual elections, and identifying controversial articles. Our analysis reveals that the interactions in co-editing articles can accurately predict votes, although there are differences between positive and negative votes. For instance, the accuracy when predicting negative votes substantially increases by considering longer traces of the edit history. As for predicting controversial articles, we show that exploiting positive and negative interactions during the production of an article provides substantial improvements on previous attempts at detecting controversial articles in Wikipedia.
@inproceedings{Rad_etaal_HT2012, author = {Hoda Sepehri-Rad and Aibek Makazhanov and Davood Rafiei and Denilson Barbosa}, title = {Leveraging editor collaboration patterns in wikipedia}, booktitle = {Proceedings of the 23rd ACM conference on Hypertext and social media}, year = {2012}, acmid = {2310001}, doi = {10.1145/2309996.2310001}, isbn = {978-1-4503-1335-3}, keywords = {admin election, controversial articles, social interactions, wikipedia}, location = {Milwaukee, Wisconsin, USA}, numpages = {10}, pages = {13--22}, publisher = {ACM}, series = {HT '12} } - Topic Classification of Blog Posts Using Distant SupervisionIn Proceedings of the Workshop on Semantic Analysis in Social Media, pp. 28–36, 2012.
Classifying blog posts by topics is useful for applications such as search and marketing. However, topic classification is time consuming and error prone, especially in an open domain such as the blogosphere. The state-of-the-art relies on supervised methods, requiring considerable training effort, that use the whole corpus vocabulary as features, demanding considerable memory to process. We show an effective alternative whereby distant supervision is used to obtain training data: we use Wikipedia articles labelled with Freebase domains. We address the memory requirements by using only named entities as features. We test our classifier on a sample of blog posts, and report up to 0.69 accuracy for multi-class labelling and 0.9 for binary classification. 1
@inproceedings{Husby_Barbosa_EACLW2012, author = {Stephanie Husby and Denilson Barbosa}, title = {Topic Classification of Blog Posts Using Distant Supervision}, booktitle = {Proceedings of the Workshop on Semantic Analysis in Social Media}, year = {2012}, address = {Avignon, France}, month = {April}, pages = {28--36}, publisher = {Association for Computational Linguistics}, url = {https://aclanthology.org/W12-0604/} }
2011
- Efficient Top-k Approximate Subtree Matching in Small MemoryIEEE Transactions on Knowledge and Data Engineering, 23, pp. 1123-1137, 2011.
We consider the Top-k Approximate Subtree Matching (tasm) problem: finding the k best matches of a small query tree within a large document tree using the canonical tree edit distance as a similarity measure between subtrees. Evaluating the tree edit distance for large XML trees is difficult: the best known algorithms have cubic runtime and quadratic space complexity, and, thus, do not scale. Our solution is tasm-postorder, a memory-efficient and scalable tasm algorithm. We prove an upper bound for the maximum subtree size for which the tree edit distance needs to be evaluated. The upper bound depends on the query and is independent of the document size and structure. A core problem is to efficiently prune subtrees that are above this size threshold. We develop an algorithm based on the prefix ring buffer that allows us to prune all subtrees above the threshold in a single postorder scan of the document. The size of the prefix ring buffer is linear in the threshold. As a result, the space complexity of tasm-postorder depends only on k and the query size, and the runtime of tasm-postorder is linear in the size of the document. Our experimental evaluation on large synthetic and real XML documents confirms our analytic results.
@article{Augsten_etal_TKDE2011, author = {Nikolaus Augsten and Denilson Barbosa and Michael M. Bohlen and Themis Palpanas}, title = {Efficient Top-k Approximate Subtree Matching in Small Memory}, journal = {IEEE Transactions on Knowledge and Data Engineering}, year = {2011}, address = {Los Alamitos, CA, USA}, doi = {10.1109/tkde.2010.245}, issn = {1041-4347}, pages = {1123-1137}, publisher = {IEEE Computer Society}, volume = {23} } - Generating Synthetic Database Schemas for Simulation PurposesIn Proceedings of the 22nd International Conference on Database and Expert Systems Applications, pp. 502–510, 2011.
@inproceedings{Pires_etal_DEXA2011, author = {Carlos Eduardo Pires and Priscilla Vieira and Marcio Saraiva and Denilson Barbosa}, title = {Generating Synthetic Database Schemas for Simulation Purposes}, booktitle = {Proceedings of the 22nd International Conference on Database and Expert Systems Applications}, year = {2011}, doi = {10.1007/978-3-642-23091-2_44}, pages = {502--510}, publisher = {Springer} } - Access Control Policy Translation, Verification, and Minimization within Heterogeneous Data FederationsACM Transactions on Information and System Security, 14(3), 2011.
Data federations provide seamless access to multiple heterogeneous and autonomous data sources pertaining to a large organization. As each source database defines its own access control policies for a set of local identities, enforcing such policies across the federation becomes a challenge. In this article, we first consider the problem of translating existing access control policies defined over source databases in a manner that allows the original semantics to be observed while becoming applicable across the entire data federation. We show that such a translation is always possible, and provide an algorithm for automating the translation. We show that verifying whether a translated policy obeys the semantics of the original access control policy defined over a source database is intractable, even under restrictive scenarios. We then describe a practical algorithmic framework for translating relational access control policies into their XML equivalent, expressed in the eXtensible Access Control Markup Language. Finally, we examine the difficulty of minimizing translated policies, and contribute a minimization algorithm applicable to nonrecursive translated policies.
@article{Leighton_Barbosa_TISSEC2011, author = {Gregory Leighton and Denilson Barbosa}, title = {Access Control Policy Translation, Verification, and Minimization within Heterogeneous Data Federations}, journal = {ACM Transactions on Information and System Security}, year = {2011}, doi = {10.1145/2043621.2043625}, number = {3}, publisher = {ACM}, volume = {14} } - Extracting Meta Statements from the BlogosphereIn Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media, pp. 225–232, 2011.
Information extraction systems have been recently proposed for organizing and exploring content in large online text corpora as information networks. In such networks, the nodes are named entities (e.g., people, organizations) while the edges correspond to statements indicating relations among such entities. To date, such systems extract rather primitive networks, capturing only those relations which are expressed by direct statements. In many applications, it is useful to also extract more subtle relations which are often expressed as meta statements in the text. These can, for instance provide the context for a statement (e.g., “Google acquired YouTube on October 2006”), or repercussion about a statement (e.g., “The US condemned Russia’s invasion of Georgia”). In this work, we report on a system for extracting relations expressed in both direct statements as well as in meta statements. We propose a method based on Conditional Random Fields that explores syntactic features to extract both kinds of statements seamlessly. We follow the Open Information Extraction paradigm, where a classifier is trained to recognize any type of relation instead of specific ones. Finally, our results show substantial improvements over a state-of-the-art information extraction system, both in terms of accuracy and, especially, recall.
@inproceedings{Mesquita_Barbosa_ICWSM2011, author = {Filipe Mesquita and Denilson Barbosa}, title = {Extracting Meta Statements from the Blogosphere}, booktitle = {Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media}, year = {2011}, doi = {10.1609/icwsm.v5i1.14135}, pages = {225--232}, publisher = {AAAI} } - Towards identifying arguments in Wikipedia pagesIn Proceedings of the 20th International Conference on World Wide Web, pp. 117-118, 2011.
Wikipedia is one of the most widely used repositories of human knowledge today, contributed mostly by a few hundred thousand regular editors. In this open environment, inevitably, differences of opinion arise among editors of the same article. Especially for polemical topics such as religion and politics, difference of opinions among editors may lead to intense "edit wars" in which editors compete to have their opinions and points of view accepted. While such disputes can compromise the reliability of the article (or at least portions of it), they are recorded in the edit history of the articles. We posit that exposing such disputes to the reader, and pointing to the portions of the text where they manifest most prominently can be beneficial in helping concerned readers in understanding such topics. In this paper, we discuss our initial efforts towards the problem of automatic evaluation of extracting controversial points in Wikipedia pages.
@inproceedings{Rad_Barbosa_WWW2011, author = {Hoda Sepehri-Rad and Denilson Barbosa}, title = {Towards identifying arguments in Wikipedia pages}, booktitle = {Proceedings of the 20th International Conference on World Wide Web}, year = {2011}, doi = {10.1145/1963192.1963252}, pages = {117-118}, publisher = {ACM} }
2010
- The Actor-Topic Model for Extracting Social Networks in Literary NarrativeIn Proceedings of the NIPS 2010 Workshop -- Machine Learning for Social Computing, pp. 7 pp, 2010.
@inproceedings{Celikyilmaz_etal_NIPS_MLSC2010, author = {Asli Celikyilmaz and Dilek Hakkani-Tur and Hua He and Greg Kondrak and Denilson Barbosa}, title = {The Actor-Topic Model for Extracting Social Networks in Literary Narrative}, booktitle = {Proceedings of the NIPS 2010 Workshop -- Machine Learning for Social Computing}, year = {2010}, pages = {7 pp} } - Analyzing natural-language artifacts of the software processIn 26th IEEE International Conference on Software Maintenance, pp. 1-5, 2010.
Software teams, as they communicate throughout the life-cycle of their projects, generate a substantial stream of textual data. Through emails and chats, developers discuss the requirements of their software system, they negotiate the distribution of tasks among them, and they make decisions about the system design, and the internal structure and functionalities of its code modules. The software research community has long recognized the importance and potential usefulness of such textual information. In this paper, we discuss our recent work on systematically analyzing several textual streams collected through our WikiDev2.0 tool. We use two different text-analysis methods to examine five different sources of textual data. We report on our experience using our method on analyzing the communications of a nine-member team over four months.
@inproceedings{Hasan_etal_ICSM2010, author = {Maryam Hasan and Eleni Stroulia and Denilson Barbosa and Manar H. Alalfi}, title = {Analyzing natural-language artifacts of the software process}, booktitle = {26th IEEE International Conference on Software Maintenance}, year = {2010}, doi = {10.1109/icsm.2010.5609680}, pages = {1-5} } - Exploring and visualizing academic social networksIn 19th ACM Conference on Information and Knowledge Management, pp. 1963-1964, 2010.
We demonstrate the ReaSoN portal, consisting of interactive web-based tools for visualizing, exploring, querying, and integrating academic social networks. We describe how these networks are automatically extracted from bibliographic and citation databases, discuss notions of visibility in such networks which enable a rich set of social network analysis, and demonstrate our novel tools for the visualization and exploration of social networks.
@inproceedings{Ganev_etal_CIKM2010, author = {Veselin Ganev and Zhaochen Guo and Diego Serrano and Denilson Barbosa and Eleni Stroulia}, title = {Exploring and visualizing academic social networks}, booktitle = {19th ACM Conference on Information and Knowledge Management}, year = {2010}, doi = {10.1145/1871437.1871786}, pages = {1963-1964}, publisher = {ACM} } - Incorporating global information into named entity recognition systems using relational contextIn Proceeding of the 33rd International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2010, pp. 883-884, 2010.
The state-of-the-art in Named Entity Recognition relies on a combination of local features of the text and global knowledge to determine the types of the recognized entities. This is problematic in some cases, resulting in entities being classified as belonging to the wrong type. We show that using global information about the corpus improves the accuracy of type identification. We explore the notion of a global domain frequency that relates relation identifying terms with pairs of entity types which are used in that relation. We use this to identify entities whose types are not compatible with the terms they co-occur in the text. Our results on a large corpus of social media content allows the identification of mistyped entities with 70% accuracy.
@inproceedings{Merhav_etal_SIGIR2010, author = {Yuval Merhav and Filipe Mesquita and Denilson Barbosa and Wai Gen Yee and Ophir Frieder}, title = {Incorporating global information into named entity recognition systems using relational context}, booktitle = {Proceeding of the 33rd International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2010}, year = {2010}, doi = {10.1145/1835449.1835664}, pages = {883-884}, publisher = {ACM} } - Access control policy translation and verification within heterogeneous data federationsIn 15th ACM Symposium on Access Control Models and Technologies, pp. 173-182, 2010.
Data federations provide seamless access to multiple heterogeneous and autonomous data sources pertaining to a large organization. As each source database defines its own access control policies for a set of local identities, enforcing such policies across the federation becomes a challenge. In this paper, we first consider the problem of translating existing access control policies defined over source databases in a manner that allows the original semantics to be observed, while becoming applicable across the entire data federation. We show that such a translation is always possible, and provide an algorithm for automating the translation. We then show that verifying that a translated policy obeys the semantics of the original access control policy defined over a source database is intractable, even under restrictive scenarios. Finally, we describe a practical algorithmic framework for translating relational access control policies into their XML equivalent, expressed in the eXtensible Access Control Markup Language.
@inproceedings{Leighton_Barbosa_SACMAT2010, author = {Gregory Leighton and Denilson Barbosa}, title = {Access control policy translation and verification within heterogeneous data federations}, booktitle = {15th ACM Symposium on Access Control Models and Technologies}, year = {2010}, doi = {10.1145/1809842.1809871}, pages = {173-182}, publisher = {ACM} } - Extracting Information Networks from the Blogosphere: State-of-the-Art and ChallengesIn 4th Int'l AAAI Conference on Weblogs and Social Media--Data Challenge, 2010.
@inproceedings{Mesquita_Barbosa_ICWSM2010, author = {Filipe Mesquita and Yuval Merhav and Denilson Barbosa}, title = {Extracting Information Networks from the Blogosphere: State-of-the-Art and Challenges}, booktitle = {4th Int'l AAAI Conference on Weblogs and Social Media--Data Challenge}, year = {2010} } - A Framework for Automatic Schema Mapping Verification Through ReasoningIn International Workshop on Data Engineering meets the Semantic Web-In conjunction with ICDE 2010, pp. 245-250, 2010.
We advocate an automated approach for verifying mappings between source and target databases in which semantics are taken into account, and that avoids two serious limitations of current verification approaches: reliance on availability of sample source and target instances, and reliance on strong statistical assumptions. We discuss how our approach can be integrated into the workflow of state-of-the-art mapping design systems, and all its necessary inputs. Our approach relies on checking the entailment of verification statements derived directly from the schema mappings and from semantic annotations to the variables used in such mappings. We discuss how such verification statements can be produced and how such annotations can be extracted from different kinds of alignments of schemas into domain ontologies. Such alignments can be derived semi-automatically; thus, our framework might prove useful in also greatly reducing the amount of input from domain experts in the development of mappings.
@inproceedings{Cappellari_etal_ICDEW2010, author = {Paolo Cappellari and Denilson Barbosa and Paolo Atzeni}, title = {A Framework for Automatic Schema Mapping Verification Through Reasoning}, booktitle = {International Workshop on Data Engineering meets the Semantic Web-In conjunction with ICDE 2010}, year = {2010}, doi = {10.1109/icdew.2010.5452703}, pages = {245-250}, publisher = {IEEE} } - TASM: Top-k Approximate Subtree MatchingIn 26th IEEE International Conference on Data Engineering, pp. 353-364, 2010.
We consider the Top-k Approximate Subtree Matching (TASM) problem: finding the k best matches of a small query tree, e.g., a DBLP article with 15 nodes, in a large document tree, e.g., DBLP with 26M nodes, using the canonical tree edit distance as a similarity measure between subtrees. Evaluating the tree edit distance for large XML trees is difficult: the best known algorithms have cubic runtime and quadratic space complexity, and, thus, do not scale. Our solution is TASM-postorder, a memory-efficient and scalable TASM algorithm. We prove an upper-bound for the maximum subtree size for which the tree edit distance needs to be evaluated. The upper bound depends on the query and is independent of the document size and structure. A core problem is to efficiently prune subtrees that are above this size threshold. We develop an algorithm based on the prefix ring buffer that allows us to prune all subtrees above the threshold in a single postorder scan of the document. The size of the prefix ring buffer is linear in the threshold. As a result, the space complexity of TASM-postorder depends only on k and the query size, and the runtime of TASM-postorder is linear in the size of the document. Our experimental evaluation on large synthetic and real XML documents confirms our analytic results.
@inproceedings{Augsten_etal_ICDE2010, author = {Nikolaus Augsten and Denilson Barbosa and Michael Bohlen and Themis Palpanas}, title = {TASM: Top-k Approximate Subtree Matching}, booktitle = {26th IEEE International Conference on Data Engineering}, year = {2010}, doi = {10.1109/icde.2010.5447905}, pages = {353-364}, publisher = {IEEE Computer Society} }
2009
- Optimizing XML CompressionIn 6th International XML Database Symposium, pp. 81-105, 2009.
The eXtensible Markup Language (XML) provides a powerful and flexible means of encoding and exchanging data. As it turns out, its main advantage as an encoding format (namely, its requirement that all open and close markup tags are present and properly balanced) yields also one of its main disadvantages: verbosity. XML-conscious compression techniques seek to overcome this drawback. Many of these techniques first separate XML structure from the document content, and then compress each independently. Further compression gains can be realized by identifying and compressing together document content that is highly similar, thereby amortizing the storage costs of auxiliary information required by the chosen compression algorithm. Additionally, the proper choice of compression algorithm is an important factor not only for the achievable compression gain, but also for access performance. Hence, choosing a compression configuration that optimizes compression gain requires one to determine (1) a partitioning strategy for document content, and (2) the best available compression algorithm to apply to each set within this partition. In this paper, we show that finding an optimal compression configuration with respect to compression gain is an NP-hard optimization problem. This problem remains intractable even if one considers a single compression algorithm for all content. We also describe an approximation algorithm for selecting a partitioning strategy for document content based on the branch-and-bound paradigm.
@inproceedings{LeightonB_XSym2009, author = {Gregory Leighton and Denilson Barbosa}, title = {Optimizing XML Compression}, booktitle = {6th International XML Database Symposium}, year = {2009}, doi = {10.1007/978-3-642-03555-5_8}, eprint = {0905.4761}, eprinttype = {arXiv}, pages = {81-105}, publisher = {Springer} } - MicrobenchmarkIn Encyclopedia of Database Systems, Springer US, pp. 1737, 2009.
@incollection{BarbosaMY_DBEnc2009a, author = {Denilson Barbosa and Ioana Manolescu and Jeffrey Xu Yu}, title = {Microbenchmark}, booktitle = {Encyclopedia of Database Systems}, year = {2009}, pages = {1737}, publisher = {Springer US} } - XML BenchmarksIn Encyclopedia of Database Systems, Springer US, pp. 3576-3579, 2009.
@incollection{BarbosaMY_DBEnc2009b, author = {Denilson Barbosa and Ioana Manolescu and Jeffrey Xu Yu}, title = {XML Benchmarks}, booktitle = {Encyclopedia of Database Systems}, year = {2009}, doi = {10.1007/978-0-387-39940-9_789}, pages = {3576-3579}, publisher = {Springer US} } - XML StorageIn Encyclopedia of Database Systems, Springer US, pp. 3627-3634, 2009.
@incollection{BarbosaBFKMVY_DBEnc09, author = {Denilson Barbosa and Philip Bohannon and Juliana Freire and Carl-Christian Kanne and Ioana Manolescu and Vasilis Vassalos and Masatoshi Yoshikawa}, title = {XML Storage}, booktitle = {Encyclopedia of Database Systems}, year = {2009}, pages = {3627-3634}, publisher = {Springer US} } - Application BenchmarkIn Encyclopedia of Database Systems, Springer US, pp. 99-100, 2009.
@incollection{BarbosaMY_DBEnc09, author = {Denilson Barbosa and Ioana Manolescu and Jeffrey Xu Yu}, title = {Application Benchmark}, booktitle = {Encyclopedia of Database Systems}, year = {2009}, crossref = {DBLP:reference/db/2009}, doi = {10.1007/978-0-387-39940-9_5055}, pages = {99-100}, publisher = {Springer US} } - An Environment for Building, Exploring and Querying Academic Social NetworksIn ACM Conference on Management of Emergent Digital Ecosystems, pp. 282–289, 2009.
Social network analysis aims at uncovering and understanding the structures and patterns resulting from social interactions among individuals and organizations engaged in a common activity. Since the early days of the field, networks are modeled as graphs modeling social actors and the relations between them. The field has become very active with the maturity of computational machinery to handle large-scale graphs, and, more recently, the automated gathering of social data. We introduce ReaSoN: a comprehensive set of tools for visualizing and exploring social networks resulting from academic research. In doing so, ReaSoN contributes to the understanding as well as fostering of the social networks underlying academic research. We describe the infrastructure, visualizations and analysis provided in our system, as well as the process of extracting the social networks which are latent in bibliographic and citation databases.
@inproceedings{GanevGSTBS_MEDES2009, author = {Veselin Ganev and Zhaochen Guo and Diego Serrano and Brendan Tansey and Denilson Barbosa and Eleni Stroulia}, title = {An Environment for Building, Exploring and Querying Academic Social Networks}, booktitle = {ACM Conference on Management of Emergent Digital Ecosystems}, year = {2009}, doi = {10.1145/1643823.1643875}, pages = {282--289}, publisher = {ACM} } - Towards Automatic Schema Mapping Verification Through ReasoningIn International Workshop on Logic in Databases, pp. 43-56, 2009.
@inproceedings{CappellariB_LID2009, author = {Paolo Cappellari and Denilson Barbosa}, title = {Towards Automatic Schema Mapping Verification Through Reasoning}, booktitle = {International Workshop on Logic in Databases}, year = {2009}, date-modified = {2015-10-01 11:52:44 +0000}, pages = {43-56} }
2007
- Database and XMLTechnologies, 5th International XML Database Symposium, XSym 2007, Vienna, Austria, September 23-24, 2007, ProceedingsXSym, Springer, 2007.
@proceedings{xsym2007, title = {Database and XMLTechnologies, 5th International XML Database Symposium, XSym 2007, Vienna, Austria, September 23-24, 2007, Proceedings}, booktitle = {XSym}, year = {2007}, doi = {DOI: 10.1007/978-3-540-75288-2}, editor = {Denilson Barbosa and Angela Bonifati and Zohra Bellahsene and Ela Hunt and Rainer Unland}, isbn = {978-3-540-75287-5}, publisher = {Springer}, series = {Lecture Notes in Computer Science}, volume = {4704} } - Labeling Data Extracted from the WebIn Proceedings of the 6th International Conference on Ontologies, DataBases, and Applications of Semantics, pp. 1099-1116, 2007.
@inproceedings{SilvaBCS_ODBASE2007, author = {Altigran Soares da Silva and Denilson Barbosa and Joao M. B. Cavalcanti and Marco A. S. Sevalho}, title = {Labeling Data Extracted from the Web}, booktitle = {Proceedings of the 6th International Conference on Ontologies, DataBases, and Applications of Semantics}, year = {2007}, doi = {10.1007/978-3-540-76848-7_72}, pages = {1099-1116}, publisher = {Springer} } - FleDEx: Flexible Data ExchangeIn Proceedings of the 9th ACM International Workshop on Web Information and Data Management, pp. 25-32, 2007.
We propose a lightweight framework for data exchange that is suitable for non-expert and casual users sharing data on the Web or through peer-to-peer systems. Unlike previous work, we consider a simplistic data model and schema formalism that are suitable for describing typical online data, and propose algorithms for mapping such schemas as well as for translating the corresponding instances. Our solution requires minimal overhead and setup costs compared to existing data exchange systems, making it very attractive in the Web data exchange setting. We report experimental results indicating that our method works well with real Web data from various domains.
@inproceedings{MesquitaBCS_WIDM2007, author = {Filipe Mesquita and Denilson Barbosa and Eli Cortez and Altigran S. da Silva}, title = {FleDEx: Flexible Data Exchange}, booktitle = {Proceedings of the 9th ACM International Workshop on Web Information and Data Management}, year = {2007}, doi = {10.1145/1316902.1316907}, pages = {25-32} } - Adaptive Record Extraction From Web PagesIn 16th World Wide Web Conference, pp. 1335–1336, 2007.
We describe an adaptive method for extracting records from web pages. Our algorithm combines a weighted tree matching metric with clustering for obtaining data extraction patterns.We compare our method experimentally to the state-of-the-art, and show that our approach is very competitive for rigidly-structured records (such as product descriptions) and far superior for loosely-structured records (such as entrieson blogs).
@inproceedings{ParkB_WWW07, author = {Justin Park and Denilson Barbosa}, title = {Adaptive Record Extraction From Web Pages}, booktitle = {16th World Wide Web Conference}, year = {2007}, address = {Banff, Alberta, Canada}, doi = {10.1145/1242572.1242838}, pages = {1335--1336}, publisher = {ACM} }
2006
- Efficient Incremental Validation of XML Documents After Composite UpdatesIn Proceedings of the Fourth International XML Database Symposium (XSym 2006), pp. 107-121, 2006.
@inproceedings{BarbosaLS_XSym2006, author = {Denilson Barbosa and Gregory Leighton and Andrew Smith}, title = {Efficient Incremental Validation of XML Documents After Composite Updates}, booktitle = {Proceedings of the Fourth International XML Database Symposium (XSym 2006)}, year = {2006}, address = {Seoul, Korea}, doi = {10.1007/11841920_8}, pages = {107-121}, publisher = {Springer} } - Declarative generation of synthetic XML dataSoftw., Pract. Exper., 36(10), pp. 1051-1079, 2006.
Synthetic data can be extremely useful in testing and evaluating algorithms, tools and systems. Most synthetic data generators available today are the result of individual benchmarking efforts. Typically, these are complex programs in which the specifications of both the structure and the contents of the data are hard‐coded. As a result, it is often difficult to customize these tools for producing synthetic data tailored for specific needs. In this article, we describe the ToXgene synthetic data generator, which is a declarative tool for generating realistic XML data for benchmarking as well as testing purposes. We present our template specification language, which consists of augmenting XML Schema with probabilistic models that guide the data‐generation process. We discuss the architecture of our current implementation and we argue about ToXgene's usefulness by discussing experimental results as well as describing two projects that use our tool. Copyright © 2006 John Wiley & Sons, Ltd.
@article{BarbosaM_SPE2006, author = {Denilson Barbosa and Alberto O. Mendelzon}, title = {Declarative generation of synthetic XML data}, journal = {Softw., Pract. Exper.}, year = {2006}, doi = {10.1002/spe.724}, number = {10}, pages = {1051-1079}, volume = {36} }
2005
- Goals and Benchmarks for Autonomic Configuration RecommendersIn Proceedings of the ACM SIGMOD International Conference on Management of Data, pp. 239-250, 2005.
We are witnessing an explosive increase in the complexity of the information systems we rely upon, Autonomic systems address this challenge by continuously configuring and tuning themselves. Recently, a number of autonomic features have been incorporated into commercial RDBMS; tools for recommending database configurations (i.e., indexes, materialized views, partitions) for a given workload are prominent examples of this promising trend.In this paper, we introduce a flexible characterization of the performance goals of configuration recommenders and develop an experimental evaluation approach to benchmark the effectiveness of these autonomic tools. We focus on exploratory queries and present extensive experimental results using both real and synthetic data that demonstrate the validity of the approach introduced. Our results identify a specific index configuration based on single-column indexes as a very useful baseline for comparisons in the exploratory setting. Furthermore, the experimental results demonstrate the unfulfilled potential for achieving improvements of several orders of magnitude.
@inproceedings{ConsensBTM_SIGMOD2005, author = {Mariano P. Consens and Denilson Barbosa and Adrian M. Teisanu and Laurent Mignet}, title = {Goals and Benchmarks for Autonomic Configuration Recommenders}, booktitle = {Proceedings of the ACM SIGMOD International Conference on Management of Data}, year = {2005}, doi = {10.1145/1066157.1066185}, pages = {239-250}, publisher = {ACM} } - Designing Information-Preserving Mapping Schemes for XMLIn Proceedings of the 31st International Conference on Very Large Data Bases, pp. 109-120, 2005.
An XML-to-relational mapping scheme consists of a procedure for shredding documents into relational databases, a procedure for publishing databases back as documents, and a set of constraints the databases must satisfy. In previous work, we defined two notions of information preservation for mapping schemes: losslessness, which guarantees that any document can be reconstructed from its corresponding database; and validation, which requires every legal database to correspond to a valid document. We also described one information-preserving mapping scheme, called Edge , and showed that, under reasonable assumptions, losslessness and validation are both undecidable. This leads to the question we study in this paper: how to design mapping schemes that are information-preserving. We propose to do it by starting with a scheme known to be information-preserving and applying to it equivalence-preserving transformations written in weakly recursive ILOG. We study an instance of this framework, the LILO algorithm, and show that it provides significant performance improvements and introduces constraints that are e#ciently enforced in practice.
@inproceedings{BarbosaFM_VLDB2005, author = {Denilson Barbosa and Juliana Freire and Alberto O. Mendelzon}, title = {Designing Information-Preserving Mapping Schemes for XML}, booktitle = {Proceedings of the 31st International Conference on Very Large Data Bases}, year = {2005}, pages = {109-120} } - Studying the XML Web: Gathering Statistics from an XML SampleWorld Wide Web, 8(4), pp. 413-438, 2005.
@article{BarbosaMV_WWWJ2005, author = {Denilson Barbosa and Laurent Mignet and Pierangelo Veltri}, title = {Studying the XML Web: Gathering Statistics from an XML Sample}, journal = {World Wide Web}, year = {2005}, doi = {10.1007/s11280-005-1544-y}, number = {4}, pages = {413-438}, volume = {8} } - Incremental Validation of XML Documents and MappingsPhD thesis, Department of Computer Science, University of Toronto, 2005.
@phdthesis{barbosa_phd05, author = {Denilson Barbosa}, title = {Incremental Validation of XML Documents and Mappings}, year = {2005}, address = {Toronto, Ontario, Canada}, school = {Department of Computer Science, University of Toronto} }
2004
- Information Preservation in XML-to-Relational MappingsIn Proceedings of the Second International XML Database Symposium (XSym 2004), pp. 66-81, 2004.
@inproceedings{BarbosaFM_XSym2004, author = {Denilson Barbosa and Juliana Freire and Alberto O. Mendelzon}, title = {Information Preservation in XML-to-Relational Mappings}, booktitle = {Proceedings of the Second International XML Database Symposium (XSym 2004)}, year = {2004}, doi = {10.1007/978-3-540-30081-6_6}, pages = {66-81} } - Efficient Incremental Validation of XML DocumentsIn Proceedings of the 20th International Conference on Data Engineering (ICDE 2004), pp. 671-682, 2004.
We discuss incremental validation of XML documents with respect to DTDs and XML schema definitions. We consider insertions and deletions of subtrees, as opposed to leaf nodes only, and we also consider the validation of ID and IDREF attributes. For arbitrary schemas, we give a worst-case n log n time and linear space algorithm, and show that it often is far superior to revalidation from scratch. We present two classes of schemas, which capture most real-life DTDs, and show that they admit a logarithmic time incremental validation algorithm that, in many cases, requires only constant auxiliary space. We then discuss an implementation of these algorithms that is independent of, and can be customized for different storage mechanisms for XML. Finally, we present extensive experimental results showing that our approach is highly efficient and scalable.
@inproceedings{BarbosaMLMA_ICDE2004, author = {Denilson Barbosa and Alberto O. Mendelzon and Leonid Libkin and Laurent Mignet and Marcelo Arenas}, title = {Efficient Incremental Validation of XML Documents}, booktitle = {Proceedings of the 20th International Conference on Data Engineering (ICDE 2004)}, year = {2004}, doi = {10.1109/ICDE.2004.1320036}, pages = {671-682} }
2003
- The XML web: a first studyIn Proceedings of the Twelfth International World Wide Web Conference, pp. 500-510, 2003.
Although originally designed for large-scale electronic publishing, XML plays an increasingly important role in the exchange of data on the Web. In fact, it is expected that XML will become the lingua franca of the Web, eventually replacing HTML. Not surprisingly, there has been a great deal of interest on XML both in industry and in academia. Nevertheless, to date no comprehensive study on the XML Web (i.e., the subset of the Web made of XML documents only) nor on its contents has been made. This paper is the first attempt at describing the XML Web and the documents contained in it. Our results are drawn from a sample of a repository of the publicly available XML documents on the Web, consisting of about 200,000 documents. Our results show that, despite its short history, XML already permeates the Web, both in terms of generic domains and geographically. Also, our results about the contents of the XML Web provide valuable input for the design of algorithms, tools and systems that use XML in one form or another.
@inproceedings{MignetBV_WWW2003, author = {Laurent Mignet and Denilson Barbosa and Pierangelo Veltri}, title = {The XML web: a first study}, booktitle = {Proceedings of the Twelfth International World Wide Web Conference}, year = {2003}, doi = {10.1145/775152.775223}, pages = {500-510}, publisher = {ACM} } - On the Characterization of the XML WebIn Proceedings of the Eleventh Italian Symposium on Advanced Database Systems (SEBD 2003), pp. 29-36, 2003.
@inproceedings{MignetBV_SEBD2003, author = {Laurent Mignet and Denilson Barbosa and Pierangelo Veltri}, title = {On the Characterization of the XML Web}, booktitle = {Proceedings of the Eleventh Italian Symposium on Advanced Database Systems (SEBD 2003)}, year = {2003}, crossref = {DBLP:conf/sebd/2003}, pages = {29-36} } - Finding ID Attributes in XML DocumentsIn Proceedings of the International XML Database Symposium, pp. 180-194, 2003.
@inproceedings{BarbosaM_XSym2003, author = {Denilson Barbosa and Alberto O. Mendelzon}, title = {Finding ID Attributes in XML Documents}, booktitle = {Proceedings of the International XML Database Symposium}, year = {2003}, crossref = {DBLP:conf/xsym/2003}, doi = {10.1007/978-3-540-39429-7_12}, pages = {180-194} }
2002
- ToXgene: An extensible template-based data generator for XMLIn Proceedings of the Fifth International Workshop on the Web and Databases (WebDB 2002), pp. 49-54, 2002.
@inproceedings{BarbosaMKL_WebDB2002, author = {Denilson Barbosa and Alberto O. Mendelzon and John Keenleyside and Kelly A. Lyons}, title = {ToXgene: An extensible template-based data generator for XML}, booktitle = {Proceedings of the Fifth International Workshop on the Web and Databases (WebDB 2002)}, year = {2002}, pages = {49-54} } - ToXgene: a template-based data generator for XMLIn Proceedings of the 2002 ACM SIGMOD International Conference on Management of Data, pp. 616, 2002.
Synthetic collections of XML documents can be useful in many applications, such as benchmarking (e.g., Xmark [4], XOO7 [2]) and algorithm testing and evaluation. We present ToXgene, a template-based tool for facilitating the generation of large, consistent collections of synthetic XML documents. ToXgene was designed with the following requirements in mind: it should be declarative, to speed the data generation up; it should be general enough to generate fairly complex XML content and it should be powerful enough to capture the most common kinds of constraints in popular benchmarks. Preliminary experimental results show that our tool can closely reproduce the data sets for the Xmark and the TPC-H benchmarks [6]. The ToXgene Template Specification Language (TSL) is a subset of the XML Schema notation augmented with annotations for specifying certain properties of the intended data, such as probability distributions, the vocabulary for CDATA content, etc. We use XML Schema as the basis for TSL not only because it is a W3C standard, but also because it provides a more detailed description of XML documents than DTDs; in particular, it allows the specification of datatypes. We note that our tool gives the user total control over the data to be generated; thus, it is intended for the cases when the user knows the structure of the data she wants and requires the data to conform to this structure (however, we note that the structure does not have to be regular). The main features of our tool are:
@inproceedings{BarbosaMKL_SIGMOD2002, author = {Denilson Barbosa and Alberto O. Mendelzon and John Keenleyside and Kelly A. Lyons}, title = {ToXgene: a template-based data generator for XML}, booktitle = {Proceedings of the 2002 ACM SIGMOD International Conference on Management of Data}, year = {2002}, doi = {10.1145/564691.564769}, pages = {616} }
2001
- ToX - the Toronto XML EngineIn Workshop on Information Integration on the Web, pp. 66-73, 2001.
@inproceedings{BarbosaBMMRR_WIIW2001, author = {Denilson Barbosa and Attila Barta and Alberto O. Mendelzon and George A. Mihaila and Flavio Rizzolo and Patricia Rodriguez-Gianolli}, title = {ToX - the Toronto XML Engine}, booktitle = {Workshop on Information Integration on the Web}, year = {2001}, pages = {66-73} }
1999
- Parallelizing MPEG Video Encoding using MultiprocessorsIn Proceedings of the XII Brazilian Symposium on Computer Graphics and Image Processing, pp. 215-222, 1999.
Many computer applications are currently using digital video. Recent advances in digital imaging and faster networking infrastructure made this technology very popular, not only for entertainment, but also in health and education applications. However, high quality video requires more storage space and communication bandwidth than traditional data. To deal with this problem, most of the digital video encoding techniques use a compression scheme. The MPEG committee has defined widely used standards for digital video encoding providing high quality images and high compression rates. However, real-time MPEG encoding also demands high computational power, usually far beyond traditional sequential computers can provide. Fortunately, the algorithms compliant to the MPEG standard can be parallelized. In this work, we propose a novel and simple shared-memory parallel algorithm for MPEG-1 video encoding based on the parallel execution of different coarse grained tasks (read pictures, write coded frames, I, P, and B frames coding). Two synchronization strategies are implemented and experimental results show that encoding rates of 43 frames per second are achieved, which is almost 50% better than the real-time encoding standard rate (30 frames per second).
@inproceedings{BarbosaKMjr_SIBGRAPI1999, author = {Denilson Barbosa and Joao Paulo Kitajima and Wagner Meira Jr.}, title = {Parallelizing MPEG Video Encoding using Multiprocessors}, booktitle = {Proceedings of the XII Brazilian Symposium on Computer Graphics and Image Processing}, year = {1999}, address = {Campinas, SP, Brazil}, doi = {10.1109/SIBGRA.1999.805727}, pages = {215-222}, publisher = {IEEE Computer Society} } - Real-time MPEG Encoding in Shared-Memory MultiprocessorsIn Proceedings of the 2nd International Conference - Parallel Computing Systems 1999, 1999.
@inproceedings{BarbosaKMJr_ICPCS1999, author = {Denilson Barbosa and João Paulo Kitajima and Wagner Meira Jr}, title = {Real-time MPEG Encoding in Shared-Memory Multiprocessors}, booktitle = {Proceedings of the 2nd International Conference - Parallel Computing Systems 1999}, year = {1999}, address = {Ensenada, Baja California, Mexico}, publisher = {CICESE} } - Compressao de Video Digital MPEG em ParaleloMSc thesis, Departamento de Ciência da Computação, Universidade Federal de Minas Gerais, 1999.
@mastersthesis{barbosa_msc1999, author = {Denilson Barbosa}, title = {Compressao de Video Digital MPEG em Paralelo}, year = {1999}, address = {Belo Horizonte, Brazil}, school = {Departamento de Ciência da Computação, Universidade Federal de Minas Gerais} }
1998
- Experimental Analysis of a Parallel Quicksort-based Algorithm for Suffix Array GenerationIn Proceedings of the 3rd International Meeting on Vector and Parallel Processing, 1998.
@inproceedings{MacedoSBCKRNNZ_VECPAR1998, author = {Autran Macedo and Elaine S. Silva and Denilson M. Barbosa and Marco A. Cristo and João Paulo Kitajima and Berthier Ribeiro-Neto and Gonzalo Navarro and Nivio Zivian}, title = {Experimental Analysis of a Parallel Quicksort-based Algorithm for Suffix Array Generation}, booktitle = {Proceedings of the 3rd International Meeting on Vector and Parallel Processing}, year = {1998}, address = {Porto, Portugal} }