Knowledge Grid for Distributed Electronic Commerce Recommendation

摘要

Traditional recommender systems are mostly centralized ones which make recommendations to users on single website by adopting the content filtering technique or collaborative filtering technique. Knowledge based systems make recommendations by knowledge reasoning and need not learn users' preferences first. There are three kinds of knowledge in knowledge based recommender systems, namely user knowledge, product knowledge and function knowledge. To give users broad choices over the ever rich information on the world wide web, grid technology is incorporated into the knowledge based recommender systems to enable the distributed electronic commerce recommendation. The hierarchical model of the knowledge grid is explored in this paper. Knowledge grid is exploited to provide a normal resource space for the constituents in the recommender systems and a uniform platform of knowledge sharing and service for recommender systems to conduct knowledge search, knowledge discovery, knowledge fusion, knowledge based workflow scheduling and execution. To realize a semantic interaction environment the semantic grid is exploited for semantic publication, discovery, integration, scheduling, allocation, execution and management of the grid resources. Ontology technology is incorporated into the knowledge grid model to annotate the heterogeneous grid resources and endow them with semantics.