Semi-supervised graph-based algorithms for word sense Disambiguation

• Grant number: 18/09465-0
• Support Opportunities: Scholarships in Brazil - Master
• Start date: December 01, 2018
• End date: January 31, 2020
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Agreement: Coordination of Improvement of Higher Education Personnel (CAPES)
• Principal Investigator: Lilian Berton
• Grantee: Samuel Bruno da Silva Sousa
• Host Institution: Instituto de Ciência e Tecnologia (ICT). Universidade Federal de São Paulo (UNIFESP). Campus São José dos Campos. São José dos Campos , SP, Brazil

Abstract

Word Sense Disambiguation (WSD) is an open problem of Natural Language Processing, which aims to identify the appropriate sense of a word in some context. Many approaches have been proposed to solve the problem, such as Knowledge-based, Supervised and Unsupervised Learning. Semi-Supervised Learning (SSL) has recently become an active research area that requires a small amount of labeled training data together with unlabeled data. In this project, we propose to employ graph-based SSL for WSD. The graph will be constructed given the senses of neighboring words. Then, a label propagation algorithm will be run on the graph-of-words to spread the sense from seed vertices to the unlabeled ones to attribute the most appropriate sense for each word. We will investigate different similarity measures for words/documents, propose new graph-of-words construction methods and analyze different label propagation algorithms. The proposed approaches will be evaluated in benchmark datasets, especially in all-words WSD task.

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)

SOUSA, SAMUEL; MILIOS, EVANGELOS; BERTON, LILIAN; IEEE. Word sense disambiguation: an evaluation study of semi-supervised approaches with word embeddings. 2020 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), v. N/A, p. 8-pg., 2020-01-01. (18/01722-3, 18/09465-0)

DUARTE, JOSE MARCIO; SOUSA, SAMUEL; MILIOS, EVANGELOS; BERTON, LILIAN. Deep analysis of word sense disambiguation via semi-supervised learning and neural word representations. INFORMATION SCIENCES, v. 570, p. 278-297, SEP 2021. (18/01722-3, 18/09465-0)