Health Social Network: a Recommender System with Heterogeneous Information Network approach

Health and health services are two inseparable parts of one's life. Each person has had different needs for health services at least several times during their life cycle and would resolve them with available facilities. Regarding the high popularity of social networks in the last two decades, one of the tools that can provide many opportunities for people in the health field is social networking. In this research, we introduce a health social network which focuses on users or patients’ association with doctors and a variety of health services. In order to improve this network’s performance, we suggest a recommender system that can offer users a doctor, a special expertise in order to ask medical consultation, or an article, based on their needs. We have used heterogeneous information networks for modeling the health social network. These networks cover several types of objects, such as physicians, patients and consultation, and also several types of relationships, such as requesting or answering a consultation. For the recommender model, we use each user’s implicit feedback which they register on the network, according to the methods provided by the heterogeneous information networks. Bayesian Personalized Ranking is used in recommender model’s learning algorithm. This algorithm is a combination of ranking scores method and the foresaid learning algorithm. In the end, we will show how to use this social network and the recommender system, by applying the suggested method on our dataset.