Earticle

초록

영어

This paper proposes a method that improves retrieval speed of an index term extracting system and shows its experimental results. This paper takes into consideration the characteristics of Korean noun syllable structures and their distributions for the higher speed of the system. For example, "꽜" ([k'at]) does not appear at the first syllable of a Korean noun. The proposed constraint on syllable structure excludes from the noun candidate list the words that contain such syllables. The distributional constraint can extract a noun stem by excluding the other forms that combine with it. This method improves processing speed because it can avoid the morphemic analyses. The proposed method is better than other ones for the analysis of sentences in the internet where many coinages and misspellings occur.

목차

 1. 들어가기  2. 앞선 연구의 검토
 3. 이름씨 추출을 위한 고려 사항
 3.1. 한국어의 특성
 3.2. 인터넷 문장의 특성
 3.3. 정확도와 속도
 4. 명사 추출
 4.1. 사전의 구성
 4.2. 명사 추출 알고리즘
 5. 실험 및 평가
 5.1. 실험
 5.2. 평가
 6. 마무리
 참고문헌

저자

• 서민정

참고문헌