Relevance Estimation in Network Information Gathering

Information Gathering is becoming a tool more and more important for network information process as well as Information Retrieval (IR) and Information Filtering (IF). In order to enhance the ratio of relevant information to the whole information gathered from network, relevance estimation must be carried out in the course of gathering. Information Gathering Systems must estimate the relevance of document content to be gathered to the query users have put forward just before actual gathering actions. We skim over TFIDF algorithm and in order to make up for its drawback, We introduce naive Bayes+KL algorithm based on the combination of naive Bayes model in Probabilistic Theory and Kullback Leibler algorithm in Information Theory. Experimental results show that this kind of probabilistic algorithm could find out more relevant documents for it comes more close to semantic comprehension.