To support the design of walking routes based on tourists’ demand, we have attempted to construct a digital feedback system. This system includes methods for creating pedestrian flow heatmaps for regional resources （discovery of potential tourist resources） and scoring recommended walking routes （quantitative evaluation of current walking routes）. As mobile devices are widely used today, the accumulation of GPS location data can be a feasible way to capture pedestrians’ city-scale movements and visualize people flows in the region at a low cost. However, recording their trajectory frequently and continuously leads to a lack of storage space at high speed and difficulties in identifying popular spots and streets from clusters in the GPS heatmap. This research extends our prior work on GPS trajectory articulation to an experimental study in the development of local tourism. This method employs various mobile sensor data to eliminate points of remaining stationary using acceleration data and abstracts indoor points with GPS horizontal accuracy data, achieving a compact and reliable dataset. Section 2 illustrates the conceptual diagram of our mobile collaborative solution：（A） designers input recommended points of interest and walking paths through the web-based editor；（B） self-guided tour applications collect walkers’ mobile sensor data and transmit their logs after the articulation process；（C） the pedestrian flow reporting system generates spot-level and path-level heatmaps with the semantic filtering （e.g., indoor staying and user content） from the dataset. The reporting system also plots the gaps between the designer’s recommended routes and the actual selection by users, called route scoring. In Section 3, using a local tourism business in Akita City as an example, we verify the superiority of our proposal by comparing （i） tracking data generated by trajectory articulation and （ii） tracking data only by GPS location data from the perspective of reliability and compactness. Finally, Section 4 reports that the introduction of our framework can provide deep insights into the improvement of walking route recommendations.
