Using graph database for file recommendation in PAD social network

In this paper, a file recommender system is introduced which is used in PAD, an academic social network at Ferdowsi University of Mashhad. Considering the large number of files shared in PAD, the proposed system is aimed at helping users in finding related files. It uses content based and collaborative filtering techniques, where the former is based on automatic tagging of file names, and the later is based on the users´ activities. Further, in order to improve efficiency, Neo4J graph database engine is employed at the data layer of the recommender system. The experimental evaluations, mainly based on the users´ feedbacks, demonstrate that the proposed system has very good performance and it provides good quality recommendations.