Research On Sequential Pattern Mining Algorithms Of Partial Order

Sequential pattern mining is an important field of data mining,it is widely applied in field of computer intrusion detection,WEB click stream analysis,DNA sequence analysis,shopping basket analysis,and many other fields.Firstly,in this thesis,the basic theories of sequential pattern mining algorithms are studied,and some related classic algorithms based on the concepts of frequent sequence and frequent closed sequence are analyzed deeply in detail.And then,basic theory of partial order pattern mining is discussed systematically,and both the related typical algorithms and their features in advantage and disadvantage are analyzed and researched under the concepts of global partial order mining and frequent closed partial order mining.Based on the concept of closed sequential pattern and the thought of TS-FCPO algorithm,a new algorithm EL-FCPO is put forward to mine equilong frequent closed partial order pattern with equal length of partial order path.The characteristic that the match between equilong sequential patterns can only be achieved to the corresponding position in a sequence is used to improve the matching method of enumeration algorithm TS-FCPO when finding matching position and to raise the mining speed.All the advantages of closed partial order method are inherited by equilong closed partial order method,and both the relationships and deterministic trends between closed sequences with equilong characteristics can be shown easily.Because it fully considers the characteristic that the match between equilong sequential patterns can only be achieved to the corresponding position in a sequence,the new method can make the practical algorithm complexity decreased obviously when it is applied in the application scenario of equilong partial order and the practical time efficiency of the algorithm is also improved.The results of this paper can be applied in formulating promotion plan of the combination of book sells online,planning route selection and strategies selection for tourists and travel companies,etc.