Recipient suggestion for electronic messages using various types of local social network data.
The purpose of this study is to design an algorithm for recommending potential recipients of messages in social networks. Social network users often want to send a message to a group of recipients. In the paper we introduce an algorithm which predicts other possible recipients given an initial set of recipients. The suggested algorithm draws upon different types of local user data: the profile, friendship graph, posts, and social interactions (likes, comments, and tags). Some of recipients of a message could be closely related and, therefore, can be viewed as a uniform group. The algorithm attempts to reveal such groups and recommends them to a user. For the purpose of evaluating the algorithm we created a demo application for Facebook. Experimental evaluation using this application demonstrated that the algorithm is able to make suggestions meaningful to experts and to improve the messenger interfaceFull text of the paper in pdf (in Russian)
Vestnik NovSU, 2014.