Building,Merging And Presenting Of Ontology Based On Formal Concept Analysis

As the foundation of the semantic web, ontology is a formal, explicit specification of a shared conceptual model. The aim of ontology is to obtain, describe and express the knowledge of related domain. Ontology provides common understanding of the domain knowledge and confirms common approbatory vocabulary in the domain, as well as gives specific definition of the relation between these vocabularies from formal model of different levels. Therefore, it becomes a pivotal issue on ontology application to build ontology. But this field is still being discussed today. There is no any mature methodology to guide ontology constructing.The construction method of ontology is mainly divided into two kinds at present: One is manual ontology building, in which the problem is: (1) In a complicated field it is time-consuming and strenuous;(2) We adopt different standards and modeling methods while building ontology, so that ontology is not currency;(3) It's of great subjectivity. It will lead to the disunity of ontology construction because different domain experts adopt different views. It is the most important that the methods which we adopt are distinct when searching concepts and relationship between concepts in domain. Furthermore, we can only find some known concepts. (from CIHAI or thesaurus).In order to reduce cost of manual ontology constructing, it is very meaningful that how to utilize technology of knowledge acquisition to obtain the ontology automatically or semi-automatically. This is the second method - Ontology Learning. By this way we can simplify the manual workload of ontology building. But the technology can't search concepts and the relation among all connotative concepts in the field, moreover, it can't present the concepts and conceptual model with definite formalized way.As a branch of applied mathematics, FCA (formal concept analysis) comes of the understanding of concept in philosophical domain. It is to describe the concept in formalization of symbol from extent and intent, and then realize the semantic information which can be understood by computer. It is to extract all connotative concepts and connections between them from formal context according to the binary relationship so as to form a hierarchical structure of concept. In fact it is a kind of concept cluster procedure to create concept lattice from data muster. Furthermore, concept lattice displays more intuitively in the form of concept lattices visualizing in three-dimensional space. We can identify and unfold sub-concept lattice in order to reducing display .According to the goal of ontology, the pivotal problem of ontology constructing is to find the concepts and relationship among concepts after confirming the field, but these are connotative in the brain or store the file of this field in actual life.The primary content of this dissertation is to apply formal concept analysis technology to obtain all connotative concepts and hierarchy of them automatically from the designated data, which is not under the influence of developer. Denoting concept in symbol achieves formalized conceptual model. The paper offers a methodology for building ontology and carries on ontology merging for knowledge sharing and reusing based on concept lattice union. Moreover, ontology is presented by expressing similar relation of concepts in non-hierarchy and the relevancy of concepts and objects in combination with probabilistic model.The main contributions are as follows:(1) Putting forward formal concept analysis method applied in ontology learning based on non-structured of source data.(2) Proposing two means of ontology presenting: 1. expressing similar relation of concepts in non-hierarchy and the relevancy of concepts and objects in combination with probabilistic model in ontology; 2. presenting ontology based on the form of concept lattices visualizing in three-dimensional space, for the sake of reducing display, proposing the determination of sub-concept lattice and the algorithm for folding and unfolding it.(3) Presenting the method of ontology merging based on formal concept analysis. According to above-mentioned theory, ontology prototype system based on FCA is designed and developed. Moreover, domain ontology is constructed by a case, which has perfect hierarchical structure of concept and semantic information to retrieval. So that search efficiency is improved.