Study On Multi-Role Social Network

The growth of social network sites is the natural result of the combination of development of Internet as the material basis and further fulfillment of socialization as the spiritual impetus. Aside from significant reflection and sufficient analysis subject of socio-economic activities, flow of information and management of social relationship, social network sites itself live in the area of social studies and computer science.This thesis focuses on the weakness of present social network sites as they take the users as actors in the network, employing a model in which actors are users'roles. By this mean, a social network is divided from a line-and-dot entirety into a three layered hierarchical structure including users, user roles, and the whole network, which is proven helpful to the issues of overflow of information, typelessness of information, lack of negotiation between senders and receivers. This is the major theoretical part.In addition, from the implemental point of view, this thesis discusses the properties of roles and an automatic role decision algorithm is proposed which utilizes the fruits of text classification, and insightfully uses the relationship with social networks among roles, enabling more specific role recommendation even when user's history is not fully prepared.To the extent, this thesis looks to improve the above-mentioned algorithm by pinpointing the refinable word list, vector classification, and role clustering. All the practical suggests are favorable to the evolvement of algorithm and the holistic research of social network sites.