A Novel Text Classification Approach Based on Enhanced Association Rule
ADMA '07 Proceedings of the 3rd international conference on Advanced Data Mining and Applications
| Hi-index | 0.00 |
Personalization recommendation is a valid method forlightening the user's burden on information retrieval. Inorder to implement personalization recommendation fortext information retrieval (TR), User-focus is defined andalgorithms for the construction of user-focus are given inthis paper. The construction of user-focus for a userdepends on his entire query requests at a period of time.Each query request of the user is treated as a transactionbetween TR system and him. Items in a transaction arenon-noise words in the query request corresponding withthe transaction. Each item is weighted with a value tomeasure the importance of the item for the user.Weighted frequent itemset is used for the user-focus'sconstruction. In order to mine weighted frequenteditemset fast, an algorithm named as WeightedFP ispresented. The experimental result shows that theimplementation of personalization recommendationbased on user-focus can lighten the user's burden causedby the work of filtering valid information from vastinformation in some degree while time requirement of TRis satisfied well.