Alexandre Alcoforado

ORCID: 0000-0003-3184-1534
Publications
Citations
Views
---
Saved
---
About
Contact & Profiles
Research Areas
  • Natural Language Processing Techniques
  • Topic Modeling
  • Hate Speech and Cyberbullying Detection
  • Multi-Agent Systems and Negotiation
  • Imbalanced Data Classification Techniques
  • Artificial Intelligence in Law
  • Speech and dialogue systems
  • Cybercrime and Law Enforcement Studies
  • Benford’s Law and Fraud Detection
  • Text and Document Classification Technologies
  • Social Media and Politics
  • Mobile Crowdsensing and Crowdsourcing
  • Comparative and International Law Studies
  • Legal Education and Practice Innovations
  • Explainable Artificial Intelligence (XAI)

Universidade de São Paulo
2021-2024

Universidade Brasil
2024

Universidade Politecnica
2021

RESUMO O uso massivo de Inteligência Artificial na moderação conteúdo internet é uma realidade dos tempos atuais. No entanto, isso levanta série questionamentos, seja sobre a pertinência do sistemas automáticos opacos, se as plataformas podem sozinhas tomar decisões que antes cabiam ao Estado. Nesse contexto, o IA “caixa-preta” passa ser considerado ameaça à liberdade expressão. Por outro lado, manter conteúdos promovam abuso virtual igualmente danoso este direito fundamental. cenário,...

10.1590/s0103-4014.202438111.020 article PT cc-by Estudos Avançados 2024-01-01

A major challenge in Natural Language Processing is obtaining annotated data for supervised learning. An option the use of crowdsourcing platforms annotation. However, introduces issues related to annotator's experience, consistency, and biases. alternative zero-shot methods, which turn have limitations compared their few-shot or fully counterparts. Recent advancements driven by large language models show potential, but struggle adapt specialized domains with severely limited data. The most...

10.48550/arxiv.2401.13229 preprint EN other-oa arXiv (Cornell University) 2024-01-01

RESUMO Um dos princípios da democracia digital é informar ativamente os cidadãos e mobilizá-los para participarem no debate político. Este artigo apresenta uma ferramenta de processamento documentos políticos públicos tornar as informações mais acessíveis aos grupos profissionais específicos. Em particular, investigamos desenvolvemos técnicas Inteligência Artificial mineração textos do Diário Assembleia República Portugal particionar, analisar, extrair sintetizar a informação das atas...

10.1590/s0103-4014.202438111.021 article PT cc-by Estudos Avançados 2024-01-01

In a context where the Brazilian judiciary system, largest in world, faces crisis due to slow processing of millions cases, it becomes imperative develop efficient methods for analyzing legal texts. We introduce uBERT, hybrid model that combines Transformer and Recurrent Neural Network architectures effectively handle long Our approach processes full text regardless its length while maintaining reasonable computational overhead. experiments demonstrate uBERT achieves superior performance...

10.48550/arxiv.2410.19184 preprint EN arXiv (Cornell University) 2024-10-24

In a context where the Brazilian judiciary system, largest in world, faces crisis due to slow processing of millions cases, it becomes imperative develop efficient methods for analyzing legal texts. We introduce uBERT, hybrid model that combines Transformer and Recurrent Neural Network architectures effectively handle long Our approach processes full text regardless its length while maintaining reasonable computational overhead. experiments demonstrate uBERT achieves superior performance...

10.5753/stil.2024.245447 article EN 2024-11-17

Subjects change frequently in moderated debates with several participants, such as parliamentary sessions, electoral debates, and trials. Partitioning a debate into blocks the same subject is essential for understanding. Often moderator responsible defining when new block begins so that task of automatically partitioning can focus solely on moderator's behavior. In this paper, we (i) propose algorithm, DEBACER, which partitions debates; (ii) carry out comparative study between conventional...

10.5753/eniac.2021.18293 preprint EN 2021-11-29

Subjects change frequently in moderated debates with several participants, such as parliamentary sessions, electoral debates, and trials. Partitioning a debate into blocks the same subject is essential for understanding. Often moderator responsible defining when new block begins so that task of automatically partitioning can focus solely on moderator's behavior. In this paper, we (i) propose algorithm, DEBACER, which partitions debates; (ii) carry out comparative study between conventional...

10.48550/arxiv.2112.05438 preprint EN other-oa arXiv (Cornell University) 2021-01-01
Coming Soon ...