• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Master's Dissertation
DOI
https://doi.org/10.11606/D.3.2022.tde-08062022-081940
Document
Author
Full name
Jose de Jesus Melendez Barros
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 2022
Supervisor
Committee
Bona, Glauber De (President)
Caseli, Helena de Medeiros
Finger, Marcelo
Title in English
A deep learning approach for aspect sentiment triplet extraction in portuguese and spanish.
Keywords in English
Aspect sentiment triplet extraction
Deep learning
Natural language processing
Abstract in English
Aspect Sentiment Triplet Extraction (ASTE) is an Aspect-Based Sentiment Analysis subtask (ABSA), which aims to extract aspect-opinion pairs from a sentence and identify the sentiment polarity associated with them. For instance, given the sentence Large rooms and great breakfast, ASTE outputs the triplet T = {(rooms, large, positive), (breakfast, great, positive)}. Although several approaches to ASBA have recently been proposed, those for Portuguese/Spanish have been mostly limited to extracting only aspects, without addressing ASTE tasks. This work aims to develop a framework based on Deep Learning to perform the Aspect Sentiment Triplet Extraction task in Portuguese and Spanish. The framework uses BERT as a context-awareness sentence encoder, multiple parallel non-linear layers to get aspect and opinion representations and a Graph Attention layer along with a Biaffine scorer to determine the sentiment dependency between each aspect-opinion pair. The comparison results show that our proposed framework significantly outperforms the baselines in Portuguese/Spanish and is competitive with its counterparts in English.
Title in Portuguese
Uma abordagem de aprendizado profundo para extração de trigêmeos de sentimento de aspecto em português e espanhol.
Keywords in Portuguese
Aprendizado computacional
Inteligência artificial
Processamento de linguagem natural
Abstract in Portuguese
Aspect Sentiment Triplet Extraction (ASTE) é uma subtarefa de Aspect-Based Sentiment Analysis (ABSA), que visa extrair pares de opiniões e aspectos de uma frase e identificar a polaridade de sentimento associada a eles. Por exemplo, dada a sentença Os quartos são amplos e ótimo café da manhã, ASTE gera o tripleto T = {(quartos, amplos, positive), (café de manhã, ótimo, positive)}. Embora várias abordagens tenham sido propostas recentemente, os trabalhos disponíveis em português e espanhol tem se limitado, em sua maioria, a extrair apenas aspectos, sem abordar tarefas ASTE. Este trabalho tem como objetivo desenvolver um framework baseado em Deep Learning para executar a tarefa de Aspect Sentiment Triplet Extraction em português e espanhol. O framework usa o BERT como um codificador de frase de consciência de contexto, várias camadas paralelas não lineares obtêm as representações dos aspectos e as opiniões e uma camada Graph Attention junto com um scorer não linear determina a dependência de sentimento entre cada par aspecto-opinião. Os resultados mostram que nosso framework proposto supera significativamente as linhas de base em português-espanhol e é competitiva com suas contrapartes em inglês.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2022-06-08
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.