Ontology-based feature-level sentiment analysis in Portuguese reviews

Abstract

Sentiment analysis is the field of study that analyses people's opinions in texts. In the last decade, humans have come to share their opinions in social media on the web. Opinions are important because whenever we need to take a decision, we want to know others' points of view. In this work we study this more detailed level of analysis, which is called the aspect-level. This more detailed type of analysis requires going deeper in the parts of a sentence and an effort to deal with their meaning, therefore knowledge rich approaches are useful. For that we use ontologies, as they can be used to represent the relevant aspects of entities, in a semantic organized way, for instance, aspects that are 'part-of' or properties of an entity in a domain of knowledge. Our approach is built for Portuguese and is evaluated on a reviews data set of the accommodation sector.