Multi-Language Sentiment Analysis for Hotel Reviews

Abstract

Touristes and traveler use avariety of information sources (e.g. travelportals, blogs, or social networking sites like twitter) to help them decide for a hotel room. These sources all contain highly subjective text that expresses the opinions of many. We took a preliminary view on user generated hotel reviews from two travel portals in English and Thai. We developed a taxonomy of features and specifically investigated how accurately they can be predicted with three classification methods. The results indicate that support vector machines perform best for this specific domain.