Emanuel Matos
From IEETA
| Subject | Extraction and Access to Information in Natural Language |  |
|---|---|---|
| Advisor | António J. S. Teixeira, Mário Rodrigues, Liliana da Silva Ferreira | |
| Group | Biomedical Informatics and Technologies | |
| Status | PhD student | |
| Starts | 2019/10/01 | |
| Ends | 2022/09/30 | |
| Country | Brazil | |
| Projects | ||
| Past projects | ||
| Events | ||
| Proposals | ||
| Personal page |
Motivation
Automatic extraction of information from these natural language sources has many applications, including Business Intelligence, Forensics, and question answering systems. Despite their potential, creation of such systems for new domains is not simple and, in general, development has been limited to developers with in-depth knowledge of the area. As extraction methods improve, the amount of information also augments, making more and more difficult effective access by humans.
Main objectives
The main objectives are: study and develop processes to simplify creation of information extraction pipelines for new domains; study and develop methods to simplify access and exploration of extracted information, exploring recent developments in question answering and dialog systems; contribute to democratization of usage of such systems in relevant economic and social areas in countries having Portuguese as official language, particularly Portugal and Brazil.
Research Questions
Initial research questions include: how to combine multiple methods and existing pipelines to improve performance and simplify development; can Open-IE be used to effectively bootstrap creation of relation extractors; how to use the extracted information to further improve extraction; how to couple extraction with users’ interface to access information.
Publications
Articles in conference proceedings
- Emanuel Matos, Mário Rodrigues, António J. S. Teixeira. Named Entity Extractors for New Domains by Transfer Learning with Automatically Annotated Data. Computational Processing of the Portuguese Language, PROPOR 2022, May 2022
- Emanuel Matos, Mário Rodrigues, Pedro Miguel, António J. S. Teixeira. Towards Automatic Creation of Annotations to Foster Development of Named Entity Recognizers. SLATE, July 2021