Tssd: Exploiting Temporal-Spatial Correlation For Service Discovery In Mobile Social Networking

Recently, the service discovery has become more and more important for intermittently-connected mobile social networking(MSN). The previous researchers focus on querying the desirable service using the label with the description of keywords. However, it is impossible for millions of services to be tagged with the accurate keywords in MSN. In this paper, we propose a non-keyword service discovery scheme for the intermittently-connected MSN. In this scheme, the temporal and spatial regularities of users are fully exploited to construct the initial community. Next, the temporal-spatial correlation community transition model is proposed to implement the time-varying community. And the service discovery is given on the basis of the temporal and spatial correlated community. Finally, we verify the relationship between the temporal and spatial factors on two well-known MSN datasets. Additionally, the performance comparisons of the proposed scheme with classical schemes show that the proposed scheme can: (a) improve the success rate of service discovery, (b) reduce the required time for service query, and (c) maintain the lower communication overhead.