• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ÇÐȸÁö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ÇÐȸÁö > µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)

Current Result Document : 5 / 8 ÀÌÀü°Ç ÀÌÀü°Ç   ´ÙÀ½°Ç ´ÙÀ½°Ç

ÇѱÛÁ¦¸ñ(Korean Title) ¸Ê¸®µà½º¸¦ ÀÌ¿ëÇÑ ±×¸®µå ±â¹Ý k-NN Á¶ÀÎ ÁúÀÇó¸® ¾Ë°í¸®Áò
¿µ¹®Á¦¸ñ(English Title) Grid-based k-Nearest Neighbor Join Query Processing Algorithm using MapReduce
ÀúÀÚ(Author) À±µé³á   Àå¹Ì¿µ   ÀåÀç¿ì   DeulNyeok Yoon   Miyoung Jang   Jae-Woo Chang  
¿ø¹®¼ö·Ïó(Citation) VOL 30 NO. 02 PP. 0103 ~ 0115 (2014. 08)
Çѱ۳»¿ë
(Korean Abstract)
ÃÖ±Ù ´ë¿ë·® µ¥ÀÌÅ͸¦ ºÐ¼®Çϱâ À§ÇÑ ¸Ê¸®µà½º ±â¹Ý ÁúÀÇó¸® ¾Ë°í¸®ÁòÀÌ ´Ù¾çÇÏ°Ô ¿¬±¸µÇ°í ÀÖ´Ù. ƯÈ÷, k-NN Á¶ÀÎ ÁúÀÇó¸® ¾Ë°í¸®ÁòÀº ¼­·Î ´Ù¸¥ µÎ °³ÀÇ µ¥ÀÌÅͺ£À̽º R°ú S°¡ Á¸ÀçÇÒ ¶§, RÀÇ ¸ðµç µ¥ÀÌÅÍ¿¡ ´ëÇØ °¡Àå °Å¸®°¡ °¡±î¿î »óÀ§ k°³ÀÇ Sµ¥ÀÌÅ͸¦ Ž»öÇÏ´Â ¾Ë°í¸®ÁòÀ¸·Î½á, µ¥ÀÌÅÍ ¸¶ÀÌ´× ¹× ºÐ¼®À» ±â¹ÝÀ¸·Î ÇÏ´Â ÀÀ¿ë ºÐ¾ß¿¡¼­ ¸Å¿ì Áß¿äÇÏ°Ô È°¿ëµÇ°í ÀÖ´Ù. ±×·¯³ª, ´ëÇ¥ ¿¬±¸ÀÎ º¸·Î³ëÀÌ ±â¹Ý k-NN Á¶ÀÎ ÁúÀÇó¸® ¾Ë°í¸®ÁòÀº º¸·Î³ëÀÌ À妽º ±¸Ãà ºñ¿ëÀÌ ¸Å¿ì Å©±â ¶§¹®¿¡, ¾÷µ¥ÀÌÆ®°¡ ºó¹øÇÏ°Ô ¹ß»ýÇÏ´Â ´ë¿ë·® µ¥ÀÌÅÍ¿¡ ÀûÇÕÇÏÁö ¸øÇÏ´Ù. ¾Æ¿ï·¯ º¸·Î³ëÀÌ ¼¿ Á¤º¸¸¦ ÀúÀåÇϱâ À§ÇØ »ç¿ëÇÏ´Â R-Æ®¸®´Â ¸Ê¸®µà½º ȯ°æ¿¡¼­ÀÇ ºÐ»ê º´·Ä 󸮿¡ ÀûÇÕÇÏÁö ¾Ê´Ù. µû¶ó¼­, º» ³í¹®¿¡¼­´Â »õ·Î¿î ±×¸®µå À妽º ±â¹ÝÀÇ k-NN Á¶ÀÎ ÁúÀÇ Ã³¸® ¾Ë°í¸®ÁòÀ» Á¦¾ÈÇÑ´Ù. ù°, ³ôÀº À妽º ±¸Ãà ºñ¿ë ¹®Á¦¸¦ ÇØ°áÇϱâ À§ÇØ, µ¥ÀÌÅÍ ºÐÆ÷¸¦ °í·ÁÇÑ µ¿Àû ±×¸®µå À妽º »ý¼º ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. µÑ°, ¸Ê¸®µà½º ȯ°æ¿¡¼­ È¿À²ÀûÀ¸·Î k-NN Á¶ÀÎ ÁúÀǸ¦ ¼öÇàÇϱâ À§ÇØ, ÀÎÁ¢¼¿ Á¤º¸¸¦ ½Ã±×´Ïó·Î È°¿ëÇÏ´Â Èĺ¸¿µ¿ª Ž»ö ¹× ÇÊÅ͸µ ¾Ë°í¸®ÁòÀ» Á¦¾ÈÇÑ´Ù. À̸¦ ÅëÇØ, RÀÇ °¢ µ¥ÀÌÅÍ°¡ À§Ä¡ÇÑ ±×¸®µå ¼¿ÀÇ ÀÎÁ¢ ¼¿¸¸À» Ž»öÇÏ¿© °ü·Ã µ¥ÀÌÅ͸¸À» ¸Ê¸®µà½ºÀÇ ÀÔ·ÂÀ¸·Î Àü¼ÛÇϱ⠶§¹®¿¡ µ¥ÀÌÅÍ ÀÔÃâ·Â ¹× ¿¬»ê ½Ã°£ÀÌ Å©°Ô °¨¼ÒÇÏ´Â ÀåÁ¡À» Áö´Ñ´Ù. ¸¶Áö¸·À¸·Î ¼º´É Æò°¡¸¦ ÅëÇØ Á¦¾ÈÇÏ´Â ±â¹ýÀÌ ³ôÀº ÁúÀÇ °á°ú Á¤È®µµ¸¦ º¸ÀÌ´Â µ¿½Ã¿¡ ÁúÀÇ Ã³¸® ½Ã°£ Ãø¸é¿¡¼­ ±âÁ¸ ±â¹ý¿¡ ºñÇØ ÃÖ´ë 3¹èÀÇ ³ôÀº ÁúÀÇ Ã³¸® ¼º´ÉÀ» ³ªÅ¸³½´Ù.
¿µ¹®³»¿ë
(English Abstract)
Recently, MapReduce based query processing algorithms have been widely studied to analyze big data. K-nearest neighbor(k-NN) join algorithm, which aims to produce the k nearest neighbors of each point of a data set S from another data set R, has been considered most important in data analysis-based applications. However, the existing k-NN join query processing algorithm suffers from high index construction cost which makes them unsuitable for big data processing. Furthermore, to store data partitioning information, the existing algorithm utilizes R-tree which is not useful in the distributed computing environment. To solve these problems, we propose a new grid-based k-NN join query processing algorithm. First, to reduce the index construction cost, we design a dynamic grid index construction algorithm by considering data distribution. Second, to efficiently perform a k-NN join query in MapReduce, we devise a candidate cell retrieval and pruning method based on data signature. Therefore, our algorithm only retrieves neighboring data from the query cell and sends them as an input of MapReduce job. This can greatly reduce the data transmission and computation overhead. In performance analysis, we prove that our algorithm outperforms the existing work up to 3 times in terms of query processing time while our algorithm achieves high query result accuracy.
Å°¿öµå(Keyword) k-NN Á¶ÀÎ ÁúÀÇó¸® ¾Ë°í¸®Áò   ¸Ê¸®µà½º ±â¹Ý ÁúÀÇ Ã³¸®   ´ë¿ë·® µ¥ÀÌÅÍ Ã³¸®   ºÐ»ê ÁúÀÇ Ã³¸®   k-NN join query processing algorithm   MapReduce-based processing   distributed data processing  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå