Research On Interactive Mining Of Association Rules

Visual data mining approaches attempt to take advantage of human abilities to perceive pattern and structure in visual form and to make sense of what are seen. It is important that the process of association rules mining are exhibited and the interaction of human and mining process is implemented.This paper has been designed and implemented an interactive visual tool of association rules mining process based on the study of association rules mining process and interactive visualization, finally makes data mining tools integrate with visualization tools. It enables the users to join in the process and control the parameters of mining process to implement the interaction of man and the process of association rules mining. All in all, the interactive visualization can improve the quality and efficiency of association rules mining and help users understand the mining results.This paper studies the concept, algorithm and process of association rules mining based on the study of data mining techniques and visual interaction techniques. The contributions of this thesis are:1. An interactive visual mining method of association rules is proposed. It is indicated that the interactive mining process of association rules should incarnate three important characteristics: the constraint of pre-process, the interaction of process and the evaluation of post-process. According to these characteristics, through the interactive mining process of association rules, the users can set different constraint conditions and parameters in different phases; it is convenient for the user to focus mining and to join in the mining process.2. A XML/EXCEL model representation method is designed and implemented. Through these models, it is convenient for rules visualization, for the mining results transferring to terminal users, and also provides a friendly interface to the related systems. 3. Interactive visual association rules learner RestrictApriori is designed and implemented and it is integrated with machine learning tools Weka. It makes up Weka's shortages which are association rules mining lacking items constraint, mining process lacking control and so on to perfect the data mining capability of Weka.