Personality-Informed Restaurant Recommendation

Recommendation systems are popular tools assisting consumers with the over-choice problem; however, they have been criticized of insufficient performance in highly complex domains. This work focuses on the analysis of consumers’ personalities, due to its recent popularity in recommender systems, within topics discussed by users in electronic word of mouth (e-WOM) to improve the recommendation of restaurants to tourists. The proposed method utilizes structured and unstructured data from online reviews to predict the probability of a user enjoying a restaurant he/she had not visited before and based on that make recommendations to different users. A personality classification model that analyses the textual information of reviews and predicts the personality of the author is employed. Topic modelling is used to identify additional features that characterize users’ preferences and restaurants features. Structured information of reviews such as restaurants’ price-range, cuisine type, and value for money are extracted and used in the prediction process. The aforementioned features are used to train an extreme gradient boosting tree model which outputs the user rating of restaurants. The trained model is compared against popular recommendation techniques such as nonnegative matrix factorization and single value decomposition.