Semantically enriched representations for Portuguese textmining: models and applications

• Grant number: 19/25010-5
• Support Opportunities: Regular Research Grants
• Duration: November 01, 2020 - April 30, 2023
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Solange Oliveira Rezende
• Grantee: Solange Oliveira Rezende
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated researchers: Alípio Mário Guedes Jorge ; Bruno Magalhães Nogueira ; Camila Vaccari Sundermann ; Marcos Aurelio Domingues ; Rafael Geraldeli Rossi ; Ricardo Marcondes Marcacini ; Roberta Akemi Sinoara ; Veronica Oliveira de Carvalho

Abstract

Text Mining techniques have become essential for supporting text analysis and knowledge discovery as the volume and variety of digital text documents have increased, either in social networks and the Web or inside organizations. Despite the application task or applied technique, the treatment of text semantics is an important challenge of the Text Mining process. The challenge is even bigger when we analyze Portuguese texts due to language particularities and the low number of Portuguese resources and researches. In this context, this project aims to advance Text Mining research, focusing on the Portuguese language, and disseminate the knowledge of the field by applying Text Mining techniques in different real-world problems. We will investigate and propose semantically enriched text representation models, considering both the vector-space model and network-based representations, as well as their application in one-class learning. As a first step to support this research, we will collect, prepare and characterize collections of texts written in Portuguese, and make consolidated information about labeled collections available to the research community. Lastly, we will evaluate and apply semantically enriched text representations in different Text Mining problems, such as sentiment analysis, recommendation systems, fake news detection, literature-based discovery and event mining. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

ARAUJO, ADAILTON F.; GOLO, MARCOS P. S.; MARCACINI, RICARDO M.. Opinion mining for app reviews: an analysis of textual representation and predictive models. AUTOMATED SOFTWARE ENGINEERING, v. 29, n. 1, MAY 2022. (19/25010-5, 19/07665-4)

SANTOS, BRUCCE NEVES DOS; MARCACINI, RICARDO MARCONDES; REZENDE, SOLANGE OLIVEIRA. Multi-Domain Aspect Extraction Using Bidirectional Encoder Representations From Transformers. IEEE ACCESS, v. 9, p. 91604-91613, 2021. (19/25010-5, 19/07665-4)

DE SOUZA, MARIANA CARAVANTI; NOGUEIRA, BRUNO MAGALHAES; ROSSI, RAFAEL GERALDELI; MARCACINI, RICARDO MARCONDES; DOS SANTOS, BRUCCE NEVES; REZENDE, SOLANGE OLIVEIRA. A network-based positive and unlabeled learning approach for fake news detection. MACHINE LEARNING, NOV 2021. (19/25010-5)