Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
¾îÅټDZâ¹Ý Ãø¸éÃßÃâ¿¡ ±â¹ÝÇÑ Á¦Ç°¸®ºäÀÇ Ãø¸é ¿ä¾à |
¿µ¹®Á¦¸ñ(English Title) |
Aspect Summarization for Product Reviews based on Attention-based Aspect Extraction |
ÀúÀÚ(Author) |
Á¤Áسç
±è»ó¿µ
±è¼ºÅÂ
ÀÌÁ¤Àç
Á¤À¯Ã¶
Jun-Nyeong Jeong
Sang-Young Kim
Seong-Tae Kim
Jeong-Jae Lee
Yuchul Jung
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 48 NO. 12 PP. 1318 ~ 1328 (2021. 12) |
Çѱ۳»¿ë (Korean Abstract) |
ÃÖ±Ù ±â°èÇнÀÀ» ÅëÇÑ ±â»ç, ³í¹® µî°ú °°Àº ¹®¼ ¿ä¾à»Ó¸¸ ¾Æ´Ï¶ó ¿Â¶óÀÎ ¸®ºä¿¡ ´ëÇÑ ¿ä¾à °ü·Ã ¿¬±¸µµ È°¹ßÇÏ´Ù. º» ¿¬±¸¿¡¼´Â ±âÁ¸ÀÇ ´Ü¼øÈ÷ ³»¿ëÀ» ¿ä¾àÇÏ´Â °Í°ú´Â ´Þ¸®, Á¦Ç° ¸®ºä¿¡ Á¸ÀçÇÏ´Â ´Ù¾çÇÑ Ãø¸é(aspect)¸¦ °í·ÁÇÏ¿© Ãø¸é ¿ä¾àÀ» »ý¼ºÇÏ´Â ±â¹ýÀ» ´Ù·é´Ù. ÇнÀ µ¥ÀÌÅÍ ±¸ÃàÀ» À§ÇØ Å©·Ñ¸µÇÑ À̾îÆù Á¦Ç°¸®ºä µ¥ÀÌÅ͸¦ Á¤Á¦ÇÏ¿© 4¸¸¿©°³ÀÇ ¸®ºä¸¦ ȹµæÇÏ¿´°í, ÀÌ Áß 4õ°³ÀÇ ¸®ºä¸¦ ¼öÀÛ¾÷À» ÅëÇØ Ãø¸é ¿ä¾à Á¤´ä ¼ÂÀ» ±¸ÃàÇÏ¿´´Ù. ƯÈ÷, Ãø¸é ±â¹Ý ´Ü¾î È®Àå ±â¹ý(ABAE)¸¦ È°¿ëÇÏ¿© ÅؽºÆ® µ¥ÀÌÅ͸¸ ÀÖÀ¸¸é Ãø¸é ¿ä¾àÀÌ °¡´ÉÇÑ ¸ðµ¨À» Á¦¾ÈÇÑ´Ù. Á¦¾È ±â¹ýÀÇ È¿À²¼ºÀ» ÆÇ´ÜÇϱâ À§ÇØ, ÇнÀ ½Ã Ãø¸é°ú °ü·ÃµÈ ´Ü¾î »ç¿ë ¿©ºÎ¿Í ¸¶½ºÅ· ºñÀ²¿¡ µû¸¥ ½ÇÇèÀ» ÁøÇàÇÏ¿´´Ù. Ãø¸é°ú °ü·ÃµÈ ´Ü¾î Áß 25%¸¦ ¹«ÀÛÀ§·Î ¸¶½ºÅ· ÇÑ ¸ðµ¨ÀÌ °¡Àå ³ôÀº ¼º´ÉÀ» º¸ÀÌ´Â °ÍÀ» È®ÀÎÇÏ¿´À¸¸ç °ËÁõ ½Ã ROUGE´Â 0.696, BERTScore´Â 0.879¸¦ ȹµæÇÏ¿´´Ù. |
¿µ¹®³»¿ë (English Abstract) |
Recently, document summaries such as articles and papers through machine learning and summary-related research on online reviews are active. In this study, unlike the existing simply summarizing content, a technique was developed for generating an aspect summary by considering various aspects of product reviews. By refining the earphone product review data crawled to build the learning data, 40,000 reviews were obtained. Moreover, we manually constructed 4,000 aspect summaries to be used for our training and evaluation tasks. In particular, we proposed a model that could summarize aspects using only text data using the aspect-based word expansion technique (ABAE). To judge the effectiveness of the proposed technique, we performed experiments according to the use of words related to aspects and the masking ratio during learning. As a result, it was confirmed that the model that randomly masked 25% of the words related to the aspect showed the highest performance, and during verification, the ROUGE was 0.696, and the BERTScore was 0.879. |
Å°¿öµå(Keyword) |
Ãø¸é ÃßÃâ
ÅؽºÆ® ¿ä¾à
Ãø¸é
¸®ºä
Æ®·£½ºÆ÷¸Ó
aspect extraction
text summarization
aspect
review
transformer
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|