A Context-Aware Mobile User Behavior Based Preference Neighbor Finding Approach for Personalized Information Retrieval

This paper adopts a new approach to searching the users with similarinterests by mobile network services instead of by desktop users. First, the direct trustis calculated based on the mobile users’ behavior andthe weight of the corresponding context (time and position), then an indirect trust degree calculation method is proposed according to the propagationdistance based on the theory of the six degrees of separation; secondly this paper further uses the evaluation score about a certain mobile web service given by a mobile user to calculate the preference similarity degree between different users. Finally based on the time attenuation functionit tries to find the users with similar preference,so as to make use of the neighbor users’ preference profile to dynamicallyupdate the target user's preference profile. Simulation shows that the proposed approach outperformspure sliding window methodand pure forgetting strategy method which fail totake the context mobile information into considerationin terms of Recall Ratio and Precision Ratio.