Computing Sentiment Analysis through Aspect-based fuzzy Aggregations

Abstract

The expression of a general opinion on a product or service may sometimes be broken down into different sub-opinions on the different aspects which characterize such a product or service. Sometimes, the general opinion about a product does not have to be the average of the other sub-opinions, but other user-dependent factors can make the aggregation of the sub-opinions more complex.This work studies some use cases with real opinions, in which fuzzy aggregation operators can help represent certain user behaviors, and hence, explain the overall assessment of an opinion with respect to partial opinions of aspects.

A T1OWA fuzzy linguistic aggregation methodology for searching feature-based opinions

“…Nevertheless, "most methods aggregate sentiment by simply averaging or taking a majority vote" [32], i.e. most existent approaches are simple numerical but not linguistic based approaches [33]. For example, in [34], the authors proposed an ontology-based framework that ranks/summarizes the main features of different products that is based on the aggregation of the detected user's preferences.…”