Personalized recommendation system based on multi_agent and rough set
Bing Wu, Fei Qing Wu, Chunming Ye · 2010
Recommendation technology is an important technology to solve problems of information overwhelmed. However, recommendation system has many shortages such as lack of ability to discover knowledge hidden in data of users' online behaviours, poor personalized service, weak recommendation efficiency and effectiveness. In this paper, we put forward a personalized recommendation system based on multi_agent. We use JADE to build a learner agent (LA) and three types of recommendation agent (RA). RA takes advantage of rough set theory to discover pattern of user's interest based on user's online behaviors. We adopt Lucene to analyse materials, which are accessed by users, and extract information from them in order to build decision table and support RA to carry on knowledge discovery work. In the meantime, we use Lucene to build personalized retrieval function to support recommendation. The system integrates advantages of various recommended methods combining with the advantages of knowledge discovery work. Experiments based on real data show that this mechanism can recommend proper learning resources to users.