A Frequent Item Sets Mining Algorithm With Constraint

With the rapidly development of database and information industry, especially along with the computer popularization, the accumulated knowledge of persons and data quantities is increasing with index, these development and progresses bring with many problems, such as data storage with enormous capacity and incensement of Large-scale database .these data need to be mined using special tools, so as to found the useful pattern. Many users hold amount of data interested in these questions.As a prospective study, data mining can solve the problems above effectively. The main research methods for data mining contain classification, clustering, series analysis, detection of associated rules, constraints and deviation analysis. The detection of associated rules is an important realm in data mining. If the Large-scale database is mined without specific aim, many problems will emerge, such as low efficiency, more redundant data and make the user more profound perplexity. The knowledge is made a servant of us and the efficiency and precision increasing when use constraints in data mining fair.This paper mainly analysis and summarize both the associated rules and constraints problem. With the above as base, a new algorithm is introduced.Firstly, in order to avoid a number of unrelated items into the algorithm process and waste time and space tremendous. Succinct constraint is used and pretreatment for item databases in this paper to gained frequent item sets which satisfied for succinct constraint, and remove redundant data to accelerate the speed of generation knowledge.Secondly, generate threshold dynamically. According to requirement of user and actual conditions, use the properties of normal distribution, generate monotonicity and anti-monotonicity threshold dynamically to constraint the data. Because the user's guidance, the results will be more precise, mining results will be more closely and the results will be interested by users because of mining is persuaded by user.Association rules aim to mine the association between items in large quantities. However, with the increment of data quantity and density, the result of data mining and utility time contradict. Therefore, the process of mining will be efficient and exact when process data by sufficient constraint and calculate dynamic threshold.