초록 close

It is important to construct Knowledge Base(like Thesaurus, Ontology, Semantic Network, etc) which can be applied to the whole field of natural language processing. For example, WordNet, Kadokawa Thesaurus, and Lexical FreeNet represent the most typical Knowledge Base in natural language processing. Many Knowledge Bases constructed in many fields does not come up to our expectations in Korean language processing. In order to construct an effective Knowledge Base, various language resources such as corpus, dictionary, synonym dictionary and WordNet have to be integrated one another, and the knowledge base has to consist of chain of morpheme-word-phrase-collocation-idiom-corpus. This paper presents a construction method and application of Korean Semantic Network (KSN). The KSN is based on Korean dictionary and Sejong corpus, and is applied to text processing, word sense disambiguation (WSD). semantic analysis, query pattern analysis in information retrieval, and so on. This paper deals with the following contents: (1) We point out problems of thesaurus and semantic network that look like a hierarchical structure of words, and compare KSN with them. The KSN has 1:1 relationship between word and sense, not 1:n relationship that an existing thesaurus and semantic network has (2) We present KSN component parts and a construction method. The KSN has noun semantic hierarchy structure linked to predicates, semantic class, proper noun, semantic information, and so on. The links are resulted from consideration of a paradigmatic relation and a syntagmatic relation within sentence. For reference, the KSN consists of dictionary, morpheme information, parts of speech information, construction information, proper noun information (name entity), noun semantic hierarchical structure, predicates classification structure, semantic class relation, idiom, semantic information, and so on. (3) We Present that WSD using the KSN is more effective than one using an existing thesaurus and semantic network.