大規模ログデータを対象とした移動経路シーケンスの類似度に基づく個人行動状況の集計と分析

本稿では，MITHRA プロジェクト (Multi-factor Identification/auTHentication ReseArch Project) のライフスタイル認証実証実験に協力頂いた数千人規模の被験者のスマートフォンから得られた数多くの滞在位置情報ログデータを対象に，被験者の個人毎の行動状況の集計と分析を実施，その結果を報告する．集計と分析にあたっては，単純な GPS (Global Positioning System) の位置情報ログデータのみを使用して，クラスタリングに基づく被験者の分類 (抽出) を行い，さらにレーベンシュタイン距離により被験者の滞在位置の経路シーケンス間の類似度を求め，これを使用して日常的な行動習慣とその傾向の分析を行った．

The Mithra Project (Multi-factor Identification / authentication ReseArch) is a research project aiming to establish technologies for creating new services using individual behavior analysis by inspecting logging data collected from personal smartphones and environmental sensing data from sensors and devices installed in various places. The project is working to establish technologies for the creation of new services using behavioral analysis, such as personal authentication and personalized daily life support based on individual behavioral characteristics and patterns. This paper reported the results of aggregation and analysis of participant individual behavior using a large amount of stay location information log data from the smartphones of thousands of contributors who cooperated in the MITHRA Project’s Lifestyle Authentication Demonstration Experiment. In the aggregation and analysis, classification and extraction of participants data were performed based on clustering technique using location log data, and their similarities of the sequences of the stay location paths of each participant were used to analyze their daily behaviors.