Collaborative filtering using evaluation values of contents from users

Abstract

When there are no evaluation values from a user who has evaluated both contents X and Z, an indirect similarity calculation unitof an arithmetic processing unitof an information processing centerindirectly calculates the similarity between the contents X and Z using evaluation values of a content Y whose evaluation value is present from a user who has evaluated both the contents X and Y and whose evaluation value is present from a user who has evaluated both the contents Y and Z. A predicted evaluation value calculation unitcalculates a predicted evaluation value from a user who has not evaluated either of the contents X and Z using the similarity between the contents X and Z calculated by the indirect similarity calculation unitand the evaluation values of the contents X and Z. Thus, it is possible to calculate the predicted evaluation values of the contents X and Z which are not directly calculable. Therefore, it becomes possible to further expand the range of contents whose evaluation values are predictable through collaborative filtering.