The Research On Mining Of Association Rules Based On RoughSet

With the widespread use of database and the rapid development of the Internet, the data stored in the database is increasing rapidly. The phenomenon of massive data failed to take full advantage of is often described as "data rich, but knowledge poor". How to mining knowledge from these massive data lead to the emergence of the field of Data Mining.The theory of RS(Rough Set),first proposed by Z.Pawlak in the early 1980s 20th century, applied to handling and extracting of vague and imprecise knowledge, aroused widespread international attention by successfully applied in the field of Data Mining etc. One of the most important application of RS in the field of Data Mining is attribution and value reduction. By reducing dimensions of attributions can sum up associate rules with decision-making supports.Today, associate rules has been applied to each field broadly. But, low efficiency, larger redundancy degree of rules relatively, concerning on just a part of associate rules by customer are common problems of mining algorithms. because of all above, RS theory combines with associate rules mining algorithms which can mine associate rules that customer interested rapidly are meaningful.Under in-depth and systematic research on RS theory and associate rules mining algorithms,This paper make some improvement based on original algorithms.The frist and the formost,this paper proposes an efficient algorithm for counting core and a reduction algorithm of attributes based on discernibility matrix which can handle the knowledge system and make the extraction of decision-rules convinent.Secondly,it put forward a mining model of association rules with decision attributes based on Apriori ,AprioriTid and AprioriHybrid algorithms, which also optimize them. Meanwhile, a approach for mining positive & negative association rules of non-redundance based on 2-level support(AMPNAR) was proposed.addtionaly,this paper proposed a approach which can constrain the frequent items and cut redundant rules down.The last but not the least,it discussed the advantages and disadvantages of standards which decided the validity of associate rules.In this paper, The ideas and the steps of Algorithm improving were described detailedly,better results were got by improved Algorithm used on experiment database and the approaches were proved valuable.