Multidimensional Hybrid Features Based Event Recommendation In EBSN

The development and popularization of online social networks has brought great convenience to people's daily life. However, because of the limited expression ability and the lack of information, this kind of online communication has also brought a few problems, such as making it more difficult to understand each other and even affecting users' mental health. With these problems, people are calling for the return of the offline face-to-face communication. In such case, EBSN (Event-Based Social Network), as a new type of heterogeneous social network, has attracted people's attention and obtained rapid development.Facing with the large number and variety of events in EBSN network platform, users often encounter event selection problem. Therefore, this paper focus on the problem of event recommendation in EBSN, which can help to solve the event selection problem and improve the quality of platform service.According to the sparseness of EBSN data, a multidimensional hybrid features based algorithm for event recommendation is proposed with full use of all kinds of information in the network. First of all, the EBSN heterogeneous network model is constructed based on the data analysis. Then, in order to mine the user preference and measure user's willingness to participate in an event, features about topology, time, spatial and semantic are extracted based on the structure and content information of the network, and a model for user's scoring on events is constructed by mixing all the features together. For the unknown weight of features in the scoring model, we design an effective algorithm to learn it. And events are recommended to user according to the prediction score of them eventually.In the paper, experiments are carried out on two real EBSN datasets using different kinds of algorithms for event recommendation. Through the comparison and analysis of the results, conclusion can be drawn that (a) our approach can effectively alleviate the sparseness problem and achieve better results, and (b) for the algorithm MHF-L which enhances the weak connection, it can achieve better effect for users with richer data, but for users with sparse data, the MHF algorithm is more appropriate. Therefore, the MHF algorithm has the best overall effect for the EBSN network with sparse users as the majority.