Towards Semantic Description of User Requirements in Recommender Systems

摘要

Nowledge based recommender systems can overcome the shortcomings of the traditional ones. To realize knowledge based recommender systems user requirements and resources on web such as information, products and services should be described as knowledge uniformly. This paper explores the semantic description of user requirements which are modeled as a triple < Goal, Constraint, Preference >. To ease the burden of knowledge expression on users an agent is adopted to elicit requirements from users. The requirements retrieval process is discussed.