A recommendation system considering users’ past/current/future contexts. idscsomumnedu

This paper proposes a recommendation system considering users ' past / current / future contexts. We dene users ' con-texts as users ' situations and conditions that inuence their preference in their information choice. The current contexts mean situations when the user just receives recommenda-tion. The past contexts mean actions the user has taken up to the present and occured situations at that time and the future contexts mean actions the user plans from now on and expected situations at that time. There are several recommendation methods based on user's action histories which is one of his/her past contexts. The methods recom-mend information based on users ' action patterns extracted from the users ' action histories. Although it is eective to use the users ' action patterns in recommendation, we thin