Ontology Learning For Chinese Language Based On Frame Semantics

With the development of information technology, the research of ontology is becoming more and more popular in the domain of computer. It plays an important part in knowledge engineering, database design and integration, information retrieval and extraction, software requirement analysis, semantic web, the analysis of object-oriented and system design based on agent. Now ontology engineering tool is relatively maturity, but construct ontology by handwork is however a fussy and a troubling task. It is likely to cause the bottleneck of knowledge access. So how to reduce the cost of ontology construction and implement the semiautomatic or automatic construction is becoming the key problem of ontology research.Ontology learning uses machine learning and statistical technologies to gain the expectation ontology from existing data source by using semiautomatic ontology-construction tools. Its main task is to extract terms, concepts and their relations from the data source. Currently, overseas direction's research is very active. One representative is the balanced cooperation modeling which was proposed by Maedche and Staab. The framework encompasses ontology import, extraction, pruning, refinement and the evaluation. Domestic study of automated ontology construction is relatively few. Shouli Li et al. profited from the overseas experience and presented an automated ontology learning technology for Chinese documents based on singular value decomposition and conceptual clustering. As a result of Chinese grammar's complexity, the study based on Chinese ontology involves natural language processing. At present most of the methods are based on the syntax analysis and the statistical method. So the sentence semantic information can not be comprehended completely. Obtainment from the corpus is mainly the syntax information.This article introduces an ontology learning technology based on Fillmore's frame semantic theory in view of the traditional ontology learning method lacking semantic information. The basic characteristic of frame semantics is the empiricism method. Its central thought is that the word significance description must relate with the semantic frame. Frame as a linguistic terminology, refers to the activated existing cognitive structure when people converse by language. This kind of cognitive structure is reflected through the words.Chapter 3 and chapter 4 introduce the frame semantics in detail. Chapter 4 first introduces the construction of Chinese frame network, the syntax analysis and semantics labeling based on frame semantics. Then compare the frame semantic labeling method's advantage with the traditional labeling method in describing the sentence semantics. Finally we select the partial semantic characteristics from the labeling sets, and form the syntax-semantics mapping structure which can be an essential factor for the later ontology learning.Chapter 5 analyzes the existing ontology learning system. We study the concepts extraction, their relations and the pattern discovered for Chinese ontology learning based on the syntax-semantics mapping structure. Also we propose an ontology learning method to make the semantic expansion for the traditional method.