A Multi-criteria Review-Based Hotel Recommendation System

Traditional Recommender Systems recommend items on the basis of a single criterion whereas Multi Criteria methods take many different criteria for each item. Although Multi Criteria Recommender Systems have a promising accuracy, approaches used by them require many previous users to first rate items with respect to these criteria. It is virtually impossible to have user ratings for every different dimension for each item. This paper presents a Multi Criteria Recommendation System for Hotel Recommendations to choose the best suited hotel in a city according to a users´ preference and other user´s reviews. In order to determine the rating of a Hotel from previous users with respect to different parameters the paper uses various Natural Language Processing approaches on a Hotel Review Corpus and builds a user-item-feature database. It also addresses the Cold Start Problem for this domain & Text Messaging Language Issue when extracting user reviews.