| 摘要 |
PROBLEM TO BE SOLVED: To recommend merchandise with highly accurate purchase certainty for the unevaluated merchandise of a user in a recommendation system for recommending merchandise at a shopping site or the like on the Internet.SOLUTION: A merchandise recommendation method by a computer includes: referring to similarity between pieces of merchandise stored in a similarity database; using an evaluation value of merchandise which has been already evaluated by a user among pieces of merchandise similar to unevaluated merchandise which has not been evaluated yet by the user; calculating an estimated evaluation value estimating the evaluation value of the unevaluated merchandise of the user and the dispersal of the estimated evaluation value with respect to a true evaluation value for each user; referring to an estimated information database in which the estimated evaluation value and the disperse are respectively associated with a merchandise name of the unevaluated merchandise; and displaying the merchandise name to a user terminal in the order of the high purchase probability of the unevaluated merchandise calculated from the purchase probability relating to the merchandise which has been already evaluated by the user who has logged in and the estimated evaluation value associated with the merchandise name of the unevaluated merchandise and the dispersal of the estimated evaluation value. |
