Opinion mining in conversational content within web discussions and commentaries

Abstract

The paper focuses on the problem of opinion classification related to web discussions and commentaries. It introduces various approaches known in this field. It also describes novelty methods, which have been designed for short conversational content processing with emphasis on dynamic analysis. This dynamic analysis is focused mainly on processing of negations and intensifiers within the opinion analysis. The contribution describes implementations of these methods for the Slovak language. The Slovak dictionaries have been created to support these implementations of lexicon based approach. In addition, the paper describes test results of the presented implementations and discussion of these results as well. © 2013 IFIP International Federation for Information Processing.