| 摘要 |
A content recommender executes a method wherein a set of attributes are provided for a plurality of content items. A number of recommendation parameters are defined as a function of attribute values for a subset of attributes and for each content item, recommendation values are determined based on the definitions. A multi-dimensional clustering is applied to the recommendation values for the content items to generate a plurality of content clusters. Each dimension of the clustering corresponds to a recommendation parameter. The content recommender then selects a set of content clusters from the content clusters and a recommendation set of content items is generated by selecting at least one content item from each selected content cluster. The invention may allow improved recommendations to be generated and may in particular allow recommendations to be generated which reflect a number of different and possibly conflicting considerations.
|
