SCIENCE@home

Sumarização contrastiva de opinião

R. Silva

Published 2019 in Unknown venue

ABSTRACT

SILVA, R. R. Contrastive opinion summarization. 2020. 151 p. Dissertação (Mestrado em Ciências – Ciências de Computação e Matemática Computacional) – Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo, São Carlos – SP, 2020. This theses presents automatic techniques for comparing opinions by generating summaries that highlight differences and similarities between two entities given a set of opinionated text. We describe and evaluate different methods for comparative opinion summarization. Three methods are brought from previous work and one is created. The input for tests consists of reviews about consumer electronic products written in Portuguese and extracted from the Web. Although there are some previously published methods, there was no study comparing them: the methods were tested on different datasets and evaluated with different metrics. Assuming that the methods will generate summaries with different characteristics for the same dataset, this paper fills this gap by building a diverse dataset and defining standardized metrics to test desirable characteristics of summaries generated by each method. Automatic summarization is important because it allows the development of tools that help users to better absorb information from a set of texts. This is especially useful if the set is too large, such as batch data collected from the Internet. Comparative opinion summarization reaches a more specific part of the problem: the case where a user wants to compare two entities based on a large volume of text that contains other people’s opinions. This research leads to a survey on how useful summaries generated by different methods are. We hypothesize that they are more effective than single-entity opinion summaries to help people understand differences between two entities. This can be beneficial for a person who wants to buy a product and is in doubt between two brands or two models. It can also be useful for a manufacturer to understand how their products rank in relation to their competitors according to popular opinion. We expect this research brings contributions both in the academic context and in the practical context. From the practical point of view, it has the potential to enable the development of tools that companies and users demand. In the academy, it will join recent research initiatives in Natural Language Processing and Opinion Mining that have gained prominence in Brazil; this project will proceed their work and bring new ideas that may be used in the future by other researchers.

PUBLICATION RECORD

CITATION MAP

EXTRACTION MAP

CLAIMS

  • No claims are published for this paper.

CONCEPTS

  • No concepts are published for this paper.

REFERENCES

Showing 1-39 of 39 references · Page 1 of 1

CITED BY

  • No citing papers are available for this paper.

Showing 0-0 of 0 citing papers · Page 1 of 1