| 摘要 |
PROBLEM TO BE SOLVED: To enable a hyper parameter to be set for each recommended user by analyzing the use tendency of the recommended user.SOLUTION: In the present invention, information of an item history database in which information of user IDs, item IDs, meta data IDs of items, and evaluation values of the items by users is stored is represented as an adjacency matrix, the adjacency matrix and recommended user IDs are received, and a hyper parameter is estimated for each recommended user ID. Using the recommended user ID, adjacency matrix, and hyper parameter, the degrees of association between the recommended user ID and items to be recommended are determined, and the item having a high degree of association is determined to be a recommended item. |
