2020³â ÄÄÇ»ÅÍÁ¾ÇÕÇмú´ëȸ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
¹®¼ °¨Á¤ ºÐ·ù¸¦ À§ÇÑ °³¼±µÈ °øµ¿ ÇнÀ Á¢±Ù |
¿µ¹®Á¦¸ñ(English Title) |
An improved co-training approach for document Sentiment classification |
ÀúÀÚ(Author) |
Huynh Cong Viet Ngu
Keon Myung Lee
À̰Ǹí
Jawad Khan
Aftab Alam
Muhammad Numan Khan
Irfan Ullah
Muhammad Umair
Umair Qudus
Tariq Habib Afridi
Sung Soo Park
Young-Koo Lee
Àڿ͵å Ä
¾ÆÇÁŸ ¹ß¶÷
¹«ÇÔ¸¶µå ´©¸¸ Ä
À̸£ÆÇ ¿ï¶ó
¹«Çϸ¶µå ¿ì¸Å¸£
±¸µÎ½º ¿ì¸Å¸£
Ÿ¸®Å© ÇϺñºê ¾ÆÇÁ ¸®µð
¹Ú¼º¼ö
ÀÌ¿µ±¸
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 47 NO. 01 PP. 0791 ~ 0793 (2020. 07) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Sentiment Analysis (SA) is an active research area that is used to automatically extract useful information from the user-generated content (UGC) to classify into positive and negative classes. Recently, various machine-learning techniques, such as supervised machine learning, semi-supervised learning, and lexicon scoring for SA have been proposed. A high-quality training data is vital to learn a sentiment classifier for textual sentiment classification, but due to various domains, the labeled data for each domain is scarce or unavailable. The manual construction of labeled corpora is a timeconsuming and laborious task because of the unstructured and unorganized nature of data. In order to address this issue, in this paper, we propose an improved co-training approach based on the n-gram and word2vec model for sentiment classification. Co-training is a semi-supervised learning approach that has effective applications in textual sentiment classification. The empirical evaluation of movie review datasets shows that the proposed approach outperforms existing techniques in terms of classification accuracy. |
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|