| 摘要 |
One example of a friend recommending apparatus according to the present invention comprises a user action analyzing unit, a database, and a friend recommending unit. The user action analyzing unit generates action attribute information for actions in each of multiple user terminals, extracts the action attribute information, which is generated in the respective user terminals during a predetermined period, according to an order of recent time to calculate the difference in time from a reference time, calculates a time weighting for a category contained in the extracted action attribute information according to the calculated time difference, calculates an interest level for the category of the extracted action attribute information by using the time weighting and a user interest level weighting, and calculates a final interest level for respective categories by adding the interest levels for the same category. The database stores the categories according to an order of magnitude of the final interest level calculated for the respective user terminals by the user action analyzing unit. Lastly, the friend recommending unit compares a category arrangement order for a friend requesting user terminal among the multiple user terminals, and a category arrangement order for a recommendation candidate user terminal by using the final interest level stored in the database, measures an interest level similarity between the interest levels of the friend requesting user terminal and the recommendation candidate user terminal, and transfers a list of friend requesting user terminals among the recommendation candidate user terminals to the friend requesting user terminal as a friend recommendation list by using the interest level similarity. |
