Exploiting Space-Time Status for Service Recommendation

The prevalence of smart devices allows people to record their spacetime status. This paper focuses on exploiting user space-time status and the related semantic information for service recommendation. Firstly, event DAG is employed to organize the space-time information generated based on the service invocation history. Generation algorithm of the event DAG is then proposed. Secondly, a novel collaborative filtering based recommendation algorithm is designed. Potentially interesting services in the target node and its subsequent nodes can be recommended. In our implementation, the user space-time status is generated from the 4D city models (3D location + time) with semantic information. A prototype system is implemented to generate service invocation logs of different users. These simulative logs are utilized to evaluate the effectiveness and efficiency of our proposed method.