Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)
ÇѱÛÁ¦¸ñ(Korean Title) |
°æ·Î ÀÓº£µù ±â¹Ý Áö½Ä ±×·¡ÇÁ ¿Ï¼º ¹æ½Ä |
¿µ¹®Á¦¸ñ(English Title) |
Path Embedding-Based Knowledge Graph Completion Approach |
ÀúÀÚ(Author) |
¹ÙÆ®¼¿·½
±è¹Î¼º
¹Ú¿µÅÃ
Batselem Jagvaral
Min-Sung Kim
Young-Tack Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 47 NO. 08 PP. 0722 ~ 0729 (2020. 08) |
Çѱ۳»¿ë (Korean Abstract) |
Áö½Ä ±×·¡ÇÁ´Â ÁúÀÇÀÀ´ä ¶Ç´Â Ãßõ½Ã½ºÅÛ°ú °°Àº Áö´ÉÇü ½Ã½ºÅÛÀ» ±¸¼ºÇϴµ¥ ¸¹ÀÌ »ç¿ëµÈ´Ù. ±×·¯³ª Áö½Ä ±×·¡ÇÁ¿¡´Â ´ëºÎºÐÀÇ ¿£Æ¼Æ¼µé »çÀÌ¿¡ °ü°è ¸µÅ©°¡ ´©¶ôµÇ¾î ÀÖ´Â ¹®Á¦°¡ Á¸ÀçÇÑ´Ù. ÀÌ·± ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ º» ³í¹®¿¡¼ BLSTM(Bidirectional LSTM) ¹× CNN(Convolutional Neural Network)À» °áÇÕÇÑ »õ·Î¿î Áö½Ä ±×·¡ÇÁ ¿Ï¼º ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ¿ì¼±, Èĺ¸ °ü°è¿Í µÎ°³ÀÇ ´ë»ó ¿£Æ¼Æ¼°¡ ÁÖ¾îÁö¸é BLSTM ¹× Convolution ¿¬»êÀ» »ç¿ëÇÏ¿© ¿£Æ¼Æ¼µéÀ» ¿¬°áÇÏ´Â °æ·ÎµéÀ» ÀúÂ÷¿ø °ø°£À¸·Î ÀÓº£µùÇÑ´Ù. ±×¸®°í ¾îÅÙ¼Ç(attention) ¸ðµ¨À» ÅëÇØ µÎ °³ÀÇ ¿£Æ¼Æ¼¸¦ Ç¥ÇöÇÏ´Â ¿©·¯ °æ·ÎµéÀ» ÇϳªÀÇ º¤ÅÍ·Î ¸¸µç´Ù. º¤ÅÍ¿Í Ãß·ÐÇÒ Èĺ¸ °ü°è »çÀÌÀÇ ¿¬°ü¼ºÀ» ÅëÇØ Èĺ¸ °ü°è°¡ ¿£Æ¼Æ¼µé°ú ¿¬°áµÉ ¼ö ÀÖ´ÂÁö¿¡ ´ëÇÑ °¡´É¼ºÀ» ¿¹ÃøÇÑ´Ù. Á¦¾ÈÇÏ´Â ¹æ¹ýÀº CNNÀ» ÀÌ¿ëÇؼ ÁÖ¾îÁø ¿£Æ¼Æ¼µéÀÇ °ü°è¸¦ Ãß·ÐÇϱ⿡ °¡Àå Áß¿äÇÑ Áö¿ª Ư¡(local feature)À» ¿£Æ¼Æ¼ »çÀÌ¿¡ ÀÖ´Â °æ·Î¿¡¼ ÃßÃâÇÏ°í BLSTMÀ» ÀÌ¿ëÇؼ ÃßÃâÇÑ Áö¿ªÆ¯Â¡ÀÇ ¼ø¼ °ü°è¿¡ ´ëÇØ ÇнÀÇÑ´Ù. À̸¦ ÅëÇØ ÀúÂ÷¿ø °æ·Î Ư¡À» È¿°úÀûÀ¸·Î ÇнÀ ÇÏ´Â °ÍÀÌ °¡´ÉÇßÀ¸¸ç, ÇнÀµÈ Ư¡µéÀ» ÀÌ¿ëÇØ ¿£Æ¼Æ¼ »çÀÌÀÇ °ü°è¸¦ ¿¹ÃøÇÏ¿´´Ù. ¿©·¯ Áö½Ä ±×·¡ÇÁ¸¦ ´ë»óÀ¸·Î ¸µÅ© ¿¹Ãø(link prediction) ½ÇÇèÀ» ÁøÇàÇßÀ¸¸ç, Á¦¾ÈÇÏ´Â ¹æ¹ýÀÌ ÃֽŠ¿¬±¸ °á°úº¸´Ù ³ôÀº ¼º´ÉÀ» º¸¿´´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
Knowledge graphs are widely used in question answering systems. However, in these circumstances most of the relations between the entities in the knowledge graph tend to be missing. To solve this issue, we propose a CNN(Convolutional Neural Network) BiLSTM(Bidirectional LSTM) based approach to infer missing links in the knowledge graphs. Our method embeds paths connecting two entities into a low-dimensional space via CNN BiLSTM. Then, an attention operation is used to attentively combine path embeddings to represent two entities. Finally, we measure the similarity between the target relation and representation of the entities to predict whether or not the relation connects those entities. By combining a CNN and BiLSTM, we are able to take advantage of the CNN¡¯s ability to recognize local patterns and the LSTM¡¯s ability to produce entity and relation ordering. In this way, it is possible to effectively identify low-dimensional path features and predict the relationships between entities using the learned features. In our experiments, we performed link prediction tasks on 4 different knowledge graphs and showed that our method achieves comparable results to state-of-the-art methods.
|
Å°¿öµå(Keyword) |
Áö½Ä ±×·¡ÇÁ ¿Ï¼º
¸µÅ© ¿¹Ãø
°æ·Î ±â¹Ý Ãß·Ð
ÀúÂ÷¿øÀÇ ÀÓº£µù
knowledge graph completion
link prediction
path-based reasoning
low-dimensional embedding
question answering
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|