Research On Recommendation System Based On Community Detection

The rapid development of information technology facilitates people's life and also leads to huge network data. How to find useful information from vast amounts of network data has been a great challenge. Link prediction is a hot issue in the field of social network analysis and link mining, and it predicts the unknown part of the network or new links in the future according to the current network. The link prediction of bipartite graph is a new direction with strong application background, which is usually used in the recommendation system.Since bipartite graphs contain edges between two types of entities, most of link prediction methods based on common neighbour applicable for unipartite graphs cannot be applied to bipartite graphs directly. In addition, there has been research to confirm that the community information is important to improve the accuracy of link prediction. However, it is neither considered comprehensively nor applied in the link prediction of bipartite graph. Therefore, it is a challenge in using community information for bipartite link prediction.The content of this paper is as follows:1 ?This paper reviews the current research on the link prediction, and focus on the link prediction task of bipartite graphs. This make a clear direction for further study on the bipartite link prediction.2? This paper proposes a bipartite link prediction method using community information named Com-BLiP. The method regards link prediction as a binary classification problem in machine learning. The local structural properties of node pair instances in a bipartite graph are extracted by the projection of the bipartite graph, together with community properties of instances by exploiting community detection techniques. Both properties are then used as vector to describe the node pair instances. The training and prediction are conducted in a supervised learning process. The experimental results in a real dataset MovieLens show that the use of community information makes the proposed method perform higher prediction accuracy and improves recommendation performance.3?Based on above research and the investigation on the requirements of a recommendation system, we design the system architecture, function module, system database for this recommendation system and finally use the Com-BLi P method to develop a real movie recommendation system based on community detection technique.