Applicability of existing anonymization methods to large location history data in urban travel

摘要

Service providers want to know user attributes and recorded information in order to improve more satisfaction of the people, or the efficiency of their services by offering services specialized to the users' preferences. However, since they choose wrong way to collect, classify, analysis, use or disclose to others, of personal information, it may exceed the explicit or implicit of the user regarding the provision of personal information. So far, many anonymization methods for those data have been proposed to solve this problem. As one of anonymous method, we focus on k-anonymization technique to realize a ‘forest from the trees’ as described above. In papers in which these methods are proposed, only qualitative analyze or examples are shown that demonstrate the usefulness of anonymized data, which are the outputs of those methods. Since it is generally said that, if the size of data gets bigger, the anonymization of data becomes easier, those methods have not been applied to real huge data. In this paper, we transform the travel records of 722,000 people traveling by train in the Tokyo area with our proposed anonymization methods, analyze the degree to which each of the results is useful, and conclude that the results are useless even when anonymity level is set to low.