An Analysis of User Behavior Similarity Based on GPS Trajectory Information

• Main Authors: Te-Chin Hsieh, 謝德瑾
• Other Authors: Jenq-Muh Hsu
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/07143919288021431426

碩士 === 國立嘉義大學 === 資訊工程學系研究所 === 100 === With the rapid development of global positioning system, users can easily obtain GPS information to recording their journey paths through GPS devices. Many researches of this issue for analyzing GPS trajectory information have been discussed. In which the analysis of user behavior similarity is worth researching direction. In this thesis, we propose the improved method for Hierarchical-Graph-based Similarity Measurement. Our proposed method considers the clustering analysis of spatial and temporal similarity. In analyzing GPS trajectory information, the feature points are mined to represent user behavior based on the stay point detection. The clustering means that stay points would be clustering in the same area through DBSCAN algorithm for simplifying and reducing the number of stay points. To compare with the cluster subsequence, user behavior similarity can be measured. Our proposed scheme can effectively enhance the analysis of the clustering points for measuring the similarity of user behaviors. The performance analysis and comparisons with the previous works are also discussed.