The Design And Implementation Of Multi-Source Recommender System Used For Traditional Chinese Medicine Search Engine

Chinese Traditional Medicine culture has a long history and also has a profundity for its variety. By visiting digital library, users can read masterpieces or other electronic records to discover the knowledge in Traditional Chinese Medicine. All these materials can be provided by using search engine exists in digital library. The search engine recommender can give us a lot of literal-related recommended words and book pages. However, the recommended words generated maybe not semantically associated and the book pages would be without sorted. Actually, readers want to learn the detail description of a query words from the recommended pages, even more, they may concern with other terminologies and other carriers of knowledge which are intrinsic related with the search term, so current search engine recommender system does not satisfy the reading cognitive model.The multi-source recommended system designed in this thesis integrates the various aspects knowledge of Chinese Traditional Medicine including medicine, prescription, illness, picture, video and etc. The aim of this system sought after is to help users learn traditional Chinese medicine knowledge more methodically and probe kinds of aspects of search items in multi-angle.First of all, the system uses auto-generated regular expression to extract information from OCR books and web. Secondly it uses Double Array Trie Tree to detach the efficacy filed of medicine and prescription, and then it uses an algorithm proposed in this thesis to treat with the components of prescription. At last it uses a dictionary learning algorithm presented in this thesis to get semantic words items prevalent with search item, combined with additional recommend items from network and logs formed as the recommended set.