Research on auto-evaluation of association rules based on information spreading theorem
基于信息扩散原理的关联规则自动评价方法研究

Regarding evaluation of association rules, existing measurements are based on even distribution of parameters in database, which can not describe the local situation of parameters. Furthermore, few parameters are used in traditional me-thods. This paper proposed five parameters for evaluating association rules as well as analysis of them at first. Then proposed partition schemes of database as well as corresponding mining algorithm. Values of parameters in different sub-database can be got by using the mining algorithm based on partitioned database. Finally, got distribution of different parameters in sub-data-base by using information spreading theorem. These kinds of distribution can be helpful to evaluation and selection of rules. Thus auto-evaluation, selection and real-time query are realized to certain extent.