On Study Of Construction Algorithm Of Concept Lattice

Nowadays, the society has entered the information era. The computer and network information technology have been developed so rapidly that data and information are increasing dramatically (information explosion) in all fields, meanwhile data and information systems become more uncertain due to human's participation. How to effectively achieve the data analysis and processing and quickly get implicit knowledge has long been an important direction of Artificial intelligence. In this background, Knowledge Discovery in Databases (KDD) and Data Mining (DM) provide a new intelligent way of understanding data.Among many methods in KDD and DM, Rough Set Theory is widely applied with its obvious advantages which do not need any preparation or additional information relevant data. Concept Lattice with complete structure and solid theory has been one of the main models in DM. For the completeness of Concept Lattices, it seems impossible to find a construction algorithm with much better time complexity than the existing algorithm. Thus, how to construct a lattice efficiently from a large context is still a key point of the research on FAC.This paper firstly discusses the development of Knowledge Discovery in Databases and Data Mining from methods to applications. The focus is the application of Rough Set and Formal Concept Analysis in KDD. The relationship between Rough set and Formal Concept Analysis is discussed. The key content is the construction algorithm of concept lattice. This chapter can be divided into two parts. The first part is about the principle of serial construction algorithm and some classical construction algorithm. A new construction algorithm based on margin concept is proposed. Comparing with the other traditional serial algorithm, the difference between them is the introduction of margin concept. After achieving margin concept, it can construct the lattice without the context. It can simultaneously construct Lattice hierarchically and generate Hasse graph. The second part of this chapter is focused on parallel algorithm which is suitable for parallel computing. With development of high-performance parallel computing, it can provide a new channel to improve algorithm in terms of time and space with the storage capacity of parallel computing. This paper also analyzes the classic ParallelNextClosure algorithm, and improves horizontal union algorithm of multiple concept lattices. The improved horizontal union algorithm can deal with the other special concept besides congener concept so far to enhance the efficiency. In the end, it proved that the horizontal union algorithm is correct and effective according to analyzing the concept lattice model of parallel structure. It also provides a new challenge of the future work.