To generate recommendation information in consideration of the browsing history of a user and a currently browsed content.
A history similarity calculating part 21 calculates similarity between a content specified by content ID and respective history contents stored in a user history management DB 40. A history evaluation value calculating part 22 calculates history similarity and a correction evaluation value EV value of the history content. A user feature vector calculating part 24 calculates the feature vector of the user based on the correction evaluation value EV value which is calculated by the history evaluation value calculating part 22. A content category evaluation value calculating part calculates a predicted value P which indicates similarity between the calculated user feature vector of the user and the feature vector of a recommendation object content to be recommended. A content transmitting part 32 rearranges the content according to the predicted value P, and determines recommendation information being the recommendation object content with respect to the user.
COPYRIGHT: (C)2011,JPO&INPIT
JP2007058398A | ||||
JP2004341584A | ||||
JP2004094383A | ||||
JP2006065650A | ||||
JP2010128927A |
WO2009125711A1 |
Yoshiki Kuroki
Takashi Okiyama
Kenji Fukaishi