Most information service today considers about location-based service or time service. However, it has more factors that make system achieve more efficient service. In this paper, we propose the relationship of time, place and purpose(TPO), preference and short time history. The three factors which time, place and purpose are mutually dependent which we have summarized the basic rules to analyze essential data. We are filtering the message with important factors are preference and history. We illustrate scenarios about "push", "pull" and "don?t disturb mode" service in Odaiba area, which our DB queries can be recommended relevant messages to users. Furthermore, we discuss the implementation of prototype system including results of experimental evaluation.