Recommendation algorithm focused on individual viewpoints

In recommender systems, it is essential to improve the adaptability not for some typical user models but for each individual. In this paper, we propose several personalization methods for content recommendation focused on individual differences in intention to select contents and in the desire to acquire information. We provide multiple content-based algorithms for TV program recommendation having respective different purposes. We then present the results of a field user test for five weeks demonstrating our new method that can enhance the adaptability for each individual, especially the method of personalizing the contribution of each attributes by extracting differences between user preference and common preference. Finally, we discuss about the necessity of dynamic algorithm adaptation depending on the characteristics of the users.