The Research Of Inter-transaction Association Rule Mining

Data Mining is a technology that obtains valid, novel, potentially useful and ultimately understandable patterns of non-trivial process from databases, data warehouses or other repositories of large amounts of data.Association rule mining,as a main task in Data Mining, the goal is to find out all concurrent relationships in the transaction project, this relationship is also known as associationThis article analysis and summary traditional inter-transaction association rules algorithm and the negative association rules algorithm,on this basis proposed a new algorithm, include the following two aspects:Firstly, the user is the object of study, model through a large number of users behavior to infer users possible behaviors by a similarity user. Propose a new algorithm used the dual-strategy model, split the source database into two kind of databases,then respectively use Markov prediction model and transaction association rules to prediction interest, so it will trim the complexity source data. reduce the workload of repeatedly scanning the database, and will also reduce the generation of useless rules, reflect the user’s interest trends more realistically, increasing the precision rate of the algorithm while improved operating efficiency of the algorithm.Secondly,For the issue of negative association rules mining, summarize existing screening negative rules, such as the way use degree of interest and related concepts,pointed out that also exist insufficient in negative association rules mining. Proposed the method of combining the correlation coefficient and chi-square test, fix the ambiguity in simply use the interest degree to measure the correlation,use the calculated results to determine whether it is positive rules, weak rules or negative rules.not only can make us get the potential information that can’t get from traditional association rule mining,but also can screen out misleading rules from traditional association rule mining.In reality, there is always a causal link between transactions, and the task of association rule is to find these links. But with the increase of the database and data redundancy, led to association rule results dramatic increased, the accuracy and the efficiency of data mining and credibility of the results of association rules mining has become more important. Adopt dual strategy divide database and conquer it by piece,use negative rules test make the mislead rules vanish,to provides a new way to association rule mining theory research.