Non-overlapping Mining Of Closed Sequential Patterns With Gap Constraints

Closed sequential pattern mining has become one of the important research in sequence pattern mining and has been widely used in many applications.Many studies have shown that the closed pattern set is lossless compression representation of the frequent pattern set,while retaining the useful information in the frequent pattern set and implying the information of other frequent patterns.Non-overlapping means that a pattern does not use the same character at any two occurrences of the same position in the sequence.Sequential pattern mining without overlapping conditions greatly reduces the generation of redundant modes and can effectively mine the modes that satisfy the conditions.Non-overlapping mining of closed sequential patterns combines the advantages of the above two mining methods,and has been widely studied and studied by more and more researchers.Compared with other types of sequential pattern mining,non-overlapping mining of closed sequential patterns is more concise and effective and the result is complete.Therefore,this paper makes an in-depth study of t non-overlapping mining of closed sequential patterns.The main research contents and related work of this article are as follows:1.We introduce the problem of non-overlapping mining of closed sequential patterns and give the relevant concepts and definitions.The mining process of related algorithms is briefly analyzed and the insufficiency of the algorithm is found out.2.The related concepts and properties of the nettree are introduced.All the occurrences of the pattern P can be converted into the path from the root node to the leaf node of the nettree,and then use the nettree structure to calculate the mode support.And the PSC algorithm based on location retrieval strategy is proposed and the non-overlapping mining of closed sequential patterns algorithms NOCSP-BEST are proposed based on this algorithm.NOCSP-BEST algorithm uses breadth-first search to generate candidate patterns.3.We carried out a large number of experiments on DNA sequences and protein sequences to analyze and compare the mining results and mining time of the above three algorithms with Clo GSgrow algorithm,and proved that NOCSP-BEST algorithms are highly efficient and complete.