지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학술대회자료
- 저자정보
- 저널정보
- 한국정보통신학회 INTERNATIONAL CONFERENCE ON FUTURE INFORMATION & COMMUNICATION ENGINEERING 2014 INTERNATIONAL CONFERENCE ON FUTURE INFORMATION & COMMUNICATION ENGINEERING Vo.6 No.1
- 발행연도
- 2014.6
- 수록면
- 483 - 486 (4page)
이용수
초록· 키워드
오류제보하기
Gini-Index has been used as a split measure for choosing the most appropriate splitting attribute in decision tree. Recently, the improved Gini-Index feature selection algorithms for text classifications based on Gini-Index theory were introduced and corrected bias for unbalanced datasets, it has proved to be better than the other methods. In addition, there were a variety of experiments for web documents, text literatures, protein functions, cardiology data, etc., however, most of them have been tested and proved by unigram features selections using the Gini-Index. Thus, in the present paper, we introduce a novel n-gram-based Gini-Index feature selection method to get n-gram representative features for text datasets. We experimented the proposed method to get relevant features for a given keyword among n-gram features and can apply to text classification, detection of topics, information retrieval, etc. The methodology, according to experimental results, could obtain more specific representative features by n-gram as well as unigram.
목차
Abstract
I. INTRODUCTION
II. NOVEL N-GRAM-BASED GINI-INDEX FEATURE SELECTION
III. EXPERIMENTAL RESULTS
IV. DISCUSSION AND CONCLUSIONS
REFERENCES
참고문헌 (0)
참고문헌 신청이 논문의 저자 정보
최근 본 자료
댓글(0)
0
UCI(KEPA) : I410-ECN-0101-2018-004-000963018