The Recognition of Multiple Virtual Identities Association Based on Multi-agent System

摘要

The recognition of multiple virtual identities association has aroused extensive attention, which can be widely used in author identification, forum spammer detection and other fields. We focus on the features of authors behavior on the dynamic data. This paper applies multi-agent system to the authors information mining fields and proposes a recognition model based on multi-agent system: MVIA-MAS. We cluster the author information in each time slice in parallel and then use association rule mining to find the target author groups, in which the multiple virtual identities are considered associated. Experiments show that the model has a better overall performance.