Topic-Oriented Search Model Based on Multi-Agent

The users´ retrieval words are distinguished, judged, and classified by utilizing the intelligence character of agent, and the concept of topic derivation is introduced. Some subtopics, which are derived from the known topic, are submitted to the Agent for searching, therefore, the retrieval results could be classified according to the topics and be convenient for user to choose. The test demonstrates that in combination the fixed topic and the topics we recommend, the knowledge warehouse is enriched for perfecting the procedure of topicderivation, the retrieval range is narrowed and the local memory is reduced.