| 摘要 |
When a certain user performs selection, information regarding a trust relationship affected by other user′s opinion, namely, having dependence on a behavior of other user, is applied to a recommendation system. A method for social recommendation according to the present invention comprises steps of mapping a user and an item to form a user-item matrix; calculating a Jaccard coefficient with respect to the user, determining that there is a social relationship when the calculated Jaccard coefficient is equal to or greater than a predetermined threshold value, and increasing a social link; and adding the increased social link to the user-item matrix. Similarity is evaluated by using a person coefficient correlation method, and the social link may be increased in further consideration of the evaluated similarity. |
