Proposal and Evaluation of Serendipitous Recommendation Method Using General Unexpectedness

systems support users in selecting items and services in an information-rich environment. Although recommender systems have been improved in terms of accuracy, such systems are still insufficient in terms of novelty and serendipity, giving unsatisfactory results to users. Two methods of "serendipitous recommendation" are therefore proposed. However, a method for recommending serendipitous items accurately to users does not yet exist, because what kinds of items are serendipitous is not clearly defined. Accordingly, a human preference model of serendipitous items based on actual data concerning a user's impression collected by questionnaires was devised. Two serendipitous recommendation methods based on the model were devised and evaluated according to a user's actual impression. The evaluation results show that one of these recommendation methods, the one using general unexpectedness independent of user profiles, can recommend the serendipitous items accurately.