Prediction of User's Special Feelings Toward Photos Based on Behaviors of Taking and Editing Photos

摘要

Recommendation of useful information is playing a fundamental role in discovering new experiences in our daily lives, and predicting a user's preferences is very significant in making services more intelligent. Additionally, due to informativeness and convenience, photos have great potential to include our thoughts, opinions, and feelings, and they seem to be suitable for extracting users' interests in realizing such intelligent services; however, even if we analyze a photo that can be used for extracting users' preferences, we cannot accurately determine that the user has an interest in an object in a photo. This study proposes a method for predicting a user's special feelings toward photos based on the user's behaviors of taking and editing photos. From our assumption, users spend considerable time taking and editing their photos when they have a special interest in an object that they are shooting. In addition, we apply our method to implement a photo sharing service to enhance people's communication by recommending photos to users that have similar behaviors. We confirm the feasibility of our proposed method by several experiments using our prototype.