Friend Recommendation Of Social Network Based On E-CARGO

With the development of information technology, the way to communicate with others becomes into online from offline. Most social networks is created by the different purposes, and their forms are diverse, among which the "SCHOLAT" is a special kind of social network for teaching and research collaboration platform which has the basic function of social network, what’s more, it mainly to provide a platform for exchange and cooperation. Because of its’ specific environment, it has certain particularity. To a certain extent, social network is a kind of mapping of real life. It not only help users to consolidate the reality of the relationship between friends, but also help users to expand their social circle. In order to enrich their own circle of friends, the network help users dig out the people who they are interested with. So friend recommendation is the one of the functions that social network must have.The research on friends’recommendation mechanism of social network is mainly to realize the friendly development of online communication. Many SCHOLATs at home and abroad have been proposed a variety way of friends’recommendation, and most of which solve the similarity between users according to the user’s personal interests or other privacy, but there is lack of specific discussion for social network domains and environments. And excessive recommendation will not only be a waste of system resources, but also cause some trouble to the user, it is not conducive to the friendly interaction between the users. Therefore, how to improve the accuracy of the recommendation and achieve the precise recommendation between users are important issues in the field of good friends’recommendation. Improvement in recommendation accuracy can not only enhance the user experience, but also is more conducive to the development of social network.This paper aims at the "SCHOLAT" which is a collaboration in teaching and research platform, and introducing the role and role relationship, combine "Role-Based Collaboration" with E-CARGO model for the network topology to build feature modeling, to match the specific cultural environment, optimization recommendation algorithm in order to enhance the user experience. Considering about the direction of the recommendation,two friend recommendation algorithm are put forward:from the perspective of topological characteristics and considering user awareness,the first method combines the topology structure of the network with hierarchical structure of model and user friend relationship of the collaborative recommendation; from the view of the matrix algebra,the second method takes user initiative as foothold and uses various matrix for modeling, also transforms the recommendation from friends problem to matrix of multi assignment problem and use the KMB algorithm to solve assignment problem between multiple users for optimal.According to the first method, this paper takes the questionnaire survey. According to the user’s recommendation method based on the topological level and recommendation direction, which proves that the algorithm is natural and friendly, and it has the promotion of user experience. By simulation the second method demonstrates the recommendation of multi assignment problem and KMB algorithm is reliable and efficient. Through precise assignment and recommendation, the algorithm improves the online social recommendation mechanism, which supports the healthy development of social network formation.