Individual Friends Recommendation Based on Random Walk with Restart in Social Networks
In social networks, current friend/user recommendation methods are mainly based on similarity measurements among users or the structure of social networks. In this paper, we design a novel friend recommendation method according to a new individual feature
- PDF / 784,481 Bytes
- 11 Pages / 439.37 x 666.142 pts Page_size
- 46 Downloads / 222 Views
2
School of Information Science and Engineering, Yanshan University, Qinhuangdao City, China [email protected], [email protected], [email protected], [email protected] The Key Laboratory for Computer Virtual Technology and System Integration of Hebei Province, Yanshan University, Qinhuangdao City, China 3 State Key Lab of Mathematical Engineering and Advanced Computing, Wuxi, China 4 Department of Systems Engineering and Engineering Management, The Chinese University of Hong Kong, Hong Kong, China [email protected]
Abstract. In social networks, current friend/user recommendation methods are mainly based on similarity measurements among users or the structure of social networks. In this paper, we design a novel friend recommendation method according to a new individual feature intimacy degree. Intimacy degree reflects the degree of interaction between two users and further indicates how close two users pay attention to each other. Specifically, we first formally define this problem and perform a theoretical investigation of the problem based on random walk with restart model. And then we design an individual friend recommendation algorithm based on the social structures and behaviors of users. At last, we conduct experiments to verify the method on a real social data set. Experimental results show that the performance of friend recommendation outperforms the existing methods, and the proposed algorithm is effective and efficient in terms of PV Value, UV Value and Conversion Rate. Keywords: Friend recommendation model · Social network analysis
1
· Intimacy degree · Random walk
Introduction
Recommendation is an effective way to reduce the cost for finding information and also a powerful way to attract customers. The flourish of the dynamic social networks provides a new environment for validating the recommendation methods, at the same time brings new challenges, e.g., how to recommend friends according to interaction information? c Springer Nature Singapore Pte Ltd. 2016 Y. Li et al. (Eds.): SMP 2016, CCIS 669, pp. 123–133, 2016. DOI: 10.1007/978-981-10-2993-6 10
124
J. Gong et al.
This paper systematically investigates the friend recommendation problem, and proposes a novel method according to a new individual feature intimacy degree. Intimacy degree is defined as the total number of reviewing, forwarding, making comments, clicking a like, replying and making @relationship which two friends give each other in social networks. As we know, interaction relation is a kind of equivalent relation and thus interaction degree can be regarded as one new feature for friend recommendation among users in social networks [1]. According to this above idea (as illustrated in Fig. 1), we propose a novel method based on the social structures and behaviors of users for friend recommendation. Specifically, we first extract friend relationships data and their interacting activities data from social networks. And then we compute the intimacy degree between target users and candidate friends using random walk algorithm (RW) base
Data Loading...
