Research Of Adaptive Text Filtering Based On Genetic Algorithm

The rapid inflation of the Internet has brought the question of the information explosion.Gaining the information of user needs accurately becomes an urgent question to solve. The technology of information filtering may help the people to pick out the interested information and shields the unnecessary information. Facing the new challenge of the real-time online network information, the technology of the adaptive information filtering come out in this case.Because of only needing the few user's needs and the servies of merits of the adaptive adjustment filtering system , adaptive information filtering becomes the the hot spot topic of the research and attention .In the text pretreatment aspect of adaptive information filtering, owing to the feature of document frequency relying on the high frequency words and mutual information relying on the low frequency word, this article has given the combination feature selection method of document frequency and mutual information and tries to eliminate their shortcoming.What's more, document frequency ratio considers document frequency in outside kind and kind. The experimental result show this method can effectively select the representative significant feature words to construct the user profile, which not only reduces the dimension of the vector space and simplify the computation, moreover reduces the redundancy of the user profile.In the adaptive study aspect of the user profile in adaptive information filtering, this article uses the course of the adaptive profile learing based on genetic algorithm for the first time.Through carring on the genetic optimization to the information of pseudo-relevance feedback of the system and choosing the most superior feature information to the Rocchio module, thus we realized the adaptive study and renewed the user profile.According to the experimental result, this method has shielded the information sparse of the pseudo-relevance feedback as well as the misleading of the feature ambiguity effectively to improve the filtering quality of the adaptive information filtering system.