Extraction of Contact-Keywords History from GPS Data and its Application for the Inference of User Preferences

Abstract

This paper proposes a method to make and analyze “contact-keywords history” from GPS logging data. We develop the method to infer gourmet preferences. The history consists of time-series attributes related to spots (shops, restaurants, etc.) in GPS logging data. These days, many contents-search or recommendation services for large volumes of contents in the Internet are available. Typically, the systems on these services need to have quite precise user requests before they can offer desirable contents. It is a hard work for a user to give enough information for the systems. Since most portable devices including mobile phones have GPS function, we can easily get GPS data and utilize them for user preferences. We evaluate the proposed method through practical GPS data.