Recommendation based on Emotional Preference

This paper describes an outline and a design of a recommender system that gives recommendations of contents based on the estimation of user’s emotional preferences. Recommender systems are essential tools in the age of information overload, and they are helpful to increase comfortability and quality of life by filtering out annoying, unnecessary and troublesome information. The standard recommender system uses personal database, which stores the profile and preferences of a user. The key point of this study is a use of emotions in the personal preferences. Emotions are subjective quantities so that are difficult to deal with. This paper explains how our system identify emotions and how to deal with them in the recommender system. We also consider risk avoidance in information presentation of recommendations.