| 摘要 |
A hybrid recommender system, in which the initial stereotype is manually defined by an expert and an affinity vector of stereotypes relating to each specific user who registers onto the system, is created to define a specific profile for each user. Recommendations for a specific user are generated according to the initial stereotype and the affinity vector of stereotypes. A binary feedback, from user regarding specific items picked by him is received (e.g., while of the item), which can be either positive or negative. Then the affinity vector of stereotypes is updated. |
