Location, time, and preference aware restaurant recommendation method

The location-based social networks introduce a platform to understand the users' preferences. In general, the users of the social networks promote the exchange of data within their own networks. Such data are being used in the literature for a wide variety of location-aware recommendation systems. We investigate the role of amalgamation of real-time GPS logs, and historical check-in data in developing a restaurant recommendation system. In this paper, we propose a novel location, time and preference aware restaurant recommendation method using the users' current geospatial location, historical check-in data of the users, and the time of the recommendation request. In the proposed method, users' check-in histories are analyzed individually to discover users' visiting trends, food preference trends, and overall popularity of the restaurants. At the same time, each restaurant's operation time and the distance are modeled separately to compute recommendation scores of the restaurants. The recommendation scores are computed by considering four key factors, namely, i) user's preference scores, ii) the distance of the restaurants, iii) the time of a day, and iv) the popularity scores of the restaurants. Each of these key factors is modeled carefully to estimate realistic recommendation scores for the restaurants in a given geospatial range. We tested our proposed method using an available online dataset. The experimental results confirm the effectiveness of the proposed method.