Knowledge Representation Theory And Applications Of The Obtained Study

With the increasing popularity of the computer and Internet, the problem between plentiful data and scanty knowledge become more and more bulge. The data in the world is increasing in surprising speed. Person in different domain is waiting for his answer from the data and exchanging information to knowledge. Therefore, it generates a new domain-Knowledge Engineering. Knowledge expressing is one of the problems in the knowledge engineering research domain. The chairman of W3C forum pointed out that knowledge expressing represented a good idea while knowledge obtaining is the key step of knowledge engineering. The main research of this paper is as follows:(1) The research of rough set theory. Mining potential information in database based on rough set theory always uses reduct to calculate all classification rules. This method may calculate too much rules and cannot satisfy the need of dynamic increasing in large database. This paper will introduce a new method, which applies the dynamic reduct not to compute all rules but stable ones. This method is admitted to decompose the universal into some subsets, which can enjoy many mutual features. So we can apply rough set method to find the reducts of these subsets. After that, we can find the stable dynamic reducts in whole table and extract valid rules. The new, unseen objects during the process of updating will be classified with these generated stable rules.(2) The research of concept lattice. Concept Lattice, also named formal concept analysis, is considered as a valid tool holding formal concept analysis. A variety of methods to abstract rules already applied widespread, especially in the database knowledge discovery. The paper will describe a valid method of mining association rules applying concept lattice. Firstly, it store the frequent item set on concept lattice. Then abstract the rules from it. At last, I show an algorithm of maintenance when database was changed.(3) Rough set and Concept Lattice. Concept lattice and rough set are efficient methods for data analysis. Those methods have been wildly applied in the field of machine studying, artificial intelligence and knowledge discovery, etc. Concept lattice and rough set in data analysis share many common characteristics. And some characters of rough set such as quivalent class, upper and under approximate all can be represented by concept lattice. In this article, the association and the irrelation between concept lattice and rough set were described. In factual application, concept lattice are wanted to be dealt with magnanimous data, so it results in too much redundancy of extracting rules immediately based upon concept lattice and brings to the waste of the computing space and the computing time of the computer. Rough set theory has the strong competence to deal with data pretreatment and distill the incertitude rules. Consequently, in order to reduce the waste of manpower and material resource, the paper proposed a method that applied rough set to construct the concept lattice, establish the general and especial model among data. The method not only improved efficiency of constructing concept lattice but also help for the efficiency of abstracting rules in concept lattice.