Opinion-meter - A Framework for Aspect-Based Sentiment Analysis

  • Darlan Santana Farias USP
  • Ivone Penque Matsuno USP
  • Ricardo Marcondes Marcacini UFMS
  • Solange Oliveira Rezende USP


With the explosion of text content made available in the internet Sentiment Analysis (SA) started to attract more of people’s attention by offering alternatives to automatically extract opinion information from text. As the internet extended its reach throughout the globe, the need for tools to enable information exchange between people who do not speak the same language emerged, to this need the most common response has been the use of Machine Translation. Some researchers have also evaluated the use of machine translation in SA and some interesting results were obtained. This work introduces Opinion-meter, a system for AspectBased Sentiment Analysis that enable users to analyze texts in several languages with the use of Machine Translation and using various methods based on PMI, Lexicon and Machine Learning. An evaluation of the methods available in the system was made in four different languages and the results suggest that although Machine Translation can yield reasonable results, Machine Learning may still be a better alternative.
Como Citar

Selecione um Formato
FARIAS, Darlan Santana; MATSUNO, Ivone Penque; MARCACINI, Ricardo Marcondes; REZENDE, Solange Oliveira. Opinion-meter - A Framework for Aspect-Based Sentiment Analysis. In: SIMPÓSIO BRASILEIRO DE SISTEMAS MULTIMÍDIA E WEB (WEBMEDIA), 22. , 2016, Teresina. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2016 . p. 351-354.