Recognizing Korean Unknown Proper Nouns by Using Automatically Extracted Lexical Clues

This paper presents a method of extracting lexical clues automatically from a very large corpus and recognizing unknown proper nouns by using those lexical clues. This method collects proper noun candidates from the raw corpus and extracts the lexical clues among the adjacent known words of the proper noun candidates. And then, it recognizes unknown nouns and determines whether the identified unknown noun is a proper noun or not by using its adjacent lexical clues. Experimental result shows that the proposed method can extract 1,416 lexical clues from about ten million word size corpus and can recognize unknown proper nouns in the test corpus in. 92% precision rate and 72% recall rate respectively.