Semi-Automatic Detection and Evaluation of Cyclist Points of Interest Using Riding Log Data Analysis

There are places that are popular with cyclists (cyclist points of interest: CPoI). However, these CPoIs are often only known within the cyclist community and are not widely shared as general tourist information. In this study, we propose a semi-automated method for detecting and evaluating CPoIs from cyclist ride log data. To verify the proposed method, we used 374 detailed ride data collected at "Michi-no-Eki Maebashi Akagi" in Maebashi City, Gunma Prefecture. By applying stop location estimation and hierarchical clustering to the collected data, we identified 125 CPoIs. We then used crowdsourcing to identify and evaluate the CPoIs. As a result, 109 locations (87.2%) matched our expectations, and a positive correlation was observed between the evaluation results. Furthermore, by analyzing the relationship between the evaluation of CPoIs and the number of visitors, we showed that it is possible to distinguish between popular CPoIs, which are widely recognized, and hidden CPoIs, which are highly rated but not well known. These results show that the proposed method is effective for detecting and evaluating CPoIs.