Research On Concept Lattice Based Association Rules Mining

As an advanced data analysis tools, and by analyzing the observed data set, data mining is used to find credible data on the relationship between the unknown data and to provide understandably, innovative and useful information to the owner. Association rule mining is to mine knowledge from a large number of data that which is valuable in describe the natural relations among items in database.The intension of each node of the concept lattice is virtually one largest item set, and it is benefit for the association rules extraction. As the central structure of the formal concept analysis, concept lattice has been used widely in software engineering, knowledge, and other areas. Because building concept lattice efficiently is an important task, incremental algorithm is obviously advantaged.Through analyzing the process of association rule mining and incrementally updating, this paper identifies the basic problems to solve. Focus on that finding frequent item sets is a key step in mining association rules, generalized list structure is employed to organize the set of concepts in concept lattice. Based on the generalized list structure, a incremental algorithm is developed to incrementally update the existing concept lattice by traversing its set of concepts with the defined visiting order. In this algorithm, the concept nodes are ordered by the numbers of their extension, we do not have to visit all nodes when we want to find the frequent nodes. Obviously, it could reduce the time complexity by this way.In the end, experimental results on artificially generated datasets are produced, which manifest that the algorithm in this paper runs much faster than the famous Apriori algorithm and Godin's algorithm.