Research Of Negative Association Rules Mining Algorithm In Web Log Mining

The Internet user behavior analysis is mainly based on Web data mining. Web data mining is to use data mining or machine learning methods to extract potentially useful patterns and information of interest to the user from the Web documents. Web data mining contains Web content mining, Web structure mining, Web usage mining (Web log mining).This paper studies the negative association rules in Web Usage Mining. First we descript the meaning of Web data mining and classification, and then explain the concept of Web usage mining and Web mining process.Secondly, we ciscuss the related concept of negative association rules, and give a brief definition and nature of the negative association rules, then leads to the concept of a strong negative association rules. After that we introduce the support of the negative association rules, the method of calculating confidence, and give a negative association rule mining steps.Thirdly, we discuss algorithm of mining the negative association rules depthly.The negative association rule mining is divided into two parts:infrequent item set mining and negative association rules mining, about which we both give in-depth discussions. In Discussion on infrequent itemsets mining by traditional mining algorithm improvements, combined with the concept of taxonomy tree this paper proposed a new infrequent itemset mining algorithms MTNE_FI_IFI; In discussing the negative association rule mining algorithm, combined with the taxonomy tree, and related degrees and CPIR of positive and negative association rule mining this paper propsed a new algorithm MINE_P_N_RULES.Fourthly, based on Web log mining theory and the theory of the negative association rules algorithm proposed in this paper, a Web log mining prototype system, which is used to mine negative association rules, is given. And then this paper descript the main function modules of the prototype system and the methods to achieve the various functional modules, and use the prototype system to mine negative association rule on the NASA-HTTP data sets. Also this paper give the implement of key functions of prototype system.Finally, this paper compared and analysised experimental results, expounded the reasons for the experimental results by the way of the theory.