Leveraging semantics for sentiment polarity detection in social media

With the increase use of microblogs and social media platforms as forms of on-line communication, we now have a huge amount of opinionated data reflecting people’s opinions and attitudes in form of reviews, forum discussions, blogs and tweets. This has recently brought great interest to sentiment analysis and opinion mining field that analyzes people’s feelings and attitudes from written language. Most of the existing approaches on sentiment analysis rely mainly on the presence of affect words that explicitly reflect sentiment. However, these approaches are semantically weak, that is, they do not take into account the semantics of words when detecting their sentiment in text. Only recently a few approaches (e.g. sentic computing) started investigating towards this direction. Following this trend, this paper investigates the role of semantics in sentiment analysis of social media. To this end, frame semantics and lexical resources such as BabelNet are employed to extract semantic features from social media that lead to more accurate sentiment analysis models. Experiments are conducted with different types of semantic information by assessing their impact in four social media datasets which incorporate tweets, blogs and movie reviews. A tenfold cross-validation shows that F1 measure increases significantly when using semantics in sentiment analysis in social media. Results show that the proposed approach considering word’s semantics for sentiment analysis surpasses non-semantic approaches for the considered datasets.