FindLoc: A Sentiment Analysis-Based Recommender System

A recommendation system (RS) is a tool for filtering information that attempts to recommend interesting items. Globally, RS is based on analyzing users’ preferences and their implicit or explicit evaluations of items. Searching for a good doctor, restaurant, school, dentist, cafeteria, etc., in real-time from the user's current location remains challenging in Algeria. In order to facilitate the finding of interesting locations for Algerian users, we propose in this work a collaborative recommendation system called FindLoc that learns to classify the different places according to their distances and previous users’ evaluations. First, we gathered users' evaluations (ratings and comments) on various locations in Guelma city. Then, a sentiment analysis method is applied to analyze the dataset and distinguish positive from negative opinions. Then, a recommendation algorithm is proposed to recommend interesting locations based on four parameters: the location (longitude and latitude), previous item ratings, users' comments, and users' ages. Real users evaluated the system, and we obtained very encouraging results in terms of precision, which was 0.91.