A Sentiment Strength Extraction Method Considering the Effect of Memory for Bicycle Navigation
Da Li, Ryuta Yamaguchi, Keisuke Ato, Tomoki Yoshihisa, Shinji Shimojo, Yukiko Kawai · 2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops) · 2022
In cities around the world, new mobility services are both welcomed and challenged by regulators and incumbent operators. For providing a better riding experience to electric bicycle users, we propose an image sentiment strength extraction method considering the effect of memory to realize the safe and comfortable route recommendation. Specifically, we apply a real time sentiment analysis API for obtaining the sentiment labels from the facial expression images as potential feedback. Then, we extract the images of road by using the coordinate on facial expression image, and provide them to users for obtaining the explicit feedback. Furthermore, we also consider the impact of memory on the explicit feedback. Thus, we send each extracted road image to users for evaluation after a period from the time of riding, and we calculate the weights as the impact of memory according to the scores of explicit feedback received at different times. Finally, we combine the weighted explicit feedback and weighted potential feedback to calculate the sentiment strength of the scenery on the route for comfortable bicycle navigation, and verify the influence of memory on sentiment strength.