Sequential Concept Lattice Model And Sequential Pattern Mining.

With the fast development of information technology, the data gain great accumulation in various fields. Knowledge discovery and data mining appear under the background, and people may take advantage of them to comprehend data. Sequence pattern mining is an important research content in knowledge discovery and data mining. Concept lattice is an effective tool which is good at describing the hierarchy relations between notions. Therefore, it is very important to research the basic theory of a concept lattice and apply the lattice theory to knowledge discovery. At present, sequence pattern mining based on the concept lattice has been a new research hotspot in data mining.This paper researches the methods for sequence pattern mining based on the concept lattice. The main work and innovations are the following:This article introduces the basic model and classical algorithms in sequence pattern mining, and discusses the applicational prospects and challenges of sequential pattern discovery. Simultaneously, the paper sums up the basic theory of concept lattice and classical algorithms for constructing lattices, and researches the basic methods and strategies for sequence pattern mining based on the concept lattice.The paper presents two new extended models for concept lattice. One is the sequence concept lattice model based on the interest measure. The model imports sequential interests to mine the sequence patterns users are interested in. The other is the sequence fuzzy concept lattice model. Firstly, it converts an exact sequence databases into a sequence fuzzy formal context. Making use of the sequence fuzzy formal context, Galois connection, sequence fuzzy conception and sequence fuzzy concept lattice are defined in detail. At last, this article presents the incremental construction algorithm of the sequence fuzzy concept lattice.Because traditional algorithms of sequence pattern mining need many scans for a database, the process costs much time and the mining result is a large set of frequent sequences that are hard to be understood for users. In order to solve these problems, this paper presents two new sequential pattern mining algorithms based on concept lattice model. One is the sequential concept lattice model based on the interest measure for mining maximal sequence patterns; the other is the sequential fuzzy concept lattice model for mining important sequential patterns. The experimental results show that algorithms can effectively mine important sequence patterns in the lattice, and have excellent performance on the time-spatial complexity.