Evaluation of Similarities of Propagation Forms on Social Network for Extracting Relationships of Information

Abstract

Information on the Internet has been increased extremely because of CGM (consumer generated media). However, such information sent by ordinary users is not managed systematically, so utilizing it is difficult especially for the beginners. To solve this issue, by identifying relationships between different information, a system should present related information for the users simultaneously. Although the existing methods such as natural language processing can evaluate the similarities of different texts, latest information or other multimedia contents, e.g., pictures, movies, etc, are difficult to evaluate. Therefore, similarities of propagation forms on a social network are proposed to utilize for extracting relationships of information. A preliminary experiment was carried out for evaluating relationships of similarity and complement can be obtained by the similarities of propagation forms of information. As a result, the messages that have many duplicate links tend to be similar, and the rate of coincidence of the directions is expected to extract complementary relationships of information.