• 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 :

ÇѱÛÁ¦¸ñ(Korean Title) ´ë¿ë·® µ¥ÀÌÅÍ¿¡¼­ µ¥ÀÌÅÍ ±ºÁýÈ­ ±â¹ÝÀÇ È¿À²ÀûÀÎ ÀÌ»óÄ¡ ŽÁö ±â¹ý
¿µ¹®Á¦¸ñ(English Title) An Efficient Outlier Detection Algorithms based on Data Clustering over Massive Data
ÀúÀÚ(Author) ±èÈ«¿¬   ¹ÎÁر⠠ Hongyeon Kim   Jun-Ki Min  
¿ø¹®¼ö·Ïó(Citation) VOL 31 NO. 03 PP. 0059 ~ 0071 (2015. 12)
Çѱ۳»¿ë
(Korean Abstract)
ÀÌ»óÄ¡¶õ µ¥ÀÌÅÍ ÁýÇÕ¿¡¼­ ´Ù¸¥ µ¥ÀÌÅ͵é°ú ºÐ¸íÇÑ Â÷À̸¦ º¸À̰ųª ¸ð¼ø °ü°è¸¦ °®´Â µ¥ÀÌÅ͸¦ ÀǹÌÇÑ´Ù. ÀÌ·¯ÇÑ Æ¯Â¡À¸·Î ÀÌ»óÄ¡´Â µ¥ÀÌÅÍ ÁýÇÕ¿¡¼­ ºÒÇÊ¿äÇÑ µ¥ÀÌÅÍ ¶Ç´Â Áß¿äÇÑ »ç°ÇÀ» ÀǹÌÇÏ´Â Á¤º¸·Î »ç¿ëµÇ±â ¶§¹®¿¡, ÀÌ»óÄ¡¸¦ ŽÁöÇÏ´Â ÀÏÀº ¸Å¿ì Áß¿äÇÏ´Ù. ±×·¯³ª ȯ°æ Á¤º¸ ÃøÁ¤°ú ¿ÜºÎ¿ÍÀÇ Åë½ÅÀÌ °¡´ÉÇÑ ±â±âµéÀÇ µîÀåÀ¸·Î µ¥ÀÌÅÍ°¡ ±âÇϱ޼öÀûÀ¸·Î ¼öÁýµÊ¿¡ µû¶ó ±âÁ¸ ÀÌ»óÄ¡ ŽÁö ±â¹ýµéÀº ÀÌ·¯ÇÑ ´ë¿ë·® µ¥ÀÌÅÍ¿¡ ÀûÇÕÇÏÁö ¾Ê´Ù. µû¶ó¼­ º» ³í¹®¿¡¼­´Â ´ë¿ë·® µ¥ÀÌÅÍ¿¡¼­ µ¥ÀÌÅÍ ±ºÁýÈ­ ±â¹ÝÀÇ È¿À²ÀûÀÎ ÀÌ»óÄ¡ ŽÁö ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. Á¦¾È ±â¹ý¿¡¼­´Â º¹¼ö °³ÀÇ ±â±âµéÀÌ Âü¿©ÇÏ´Â ¸Ê¸®µà½º¸¦ ÀÌ¿ëÇÏ¿© Áö¿ª ±ºÁýµéÀ» º´·Ä·Î »ý¼ºÇÏ°í, Áö¿ª ±ºÁýµéÀ» º´ÇÕÇÏ¿© Àü¿ª ±ºÁýµé·Î »ý¼ºÇÑ´Ù. Á¦¾È ±â¹ý¿¡¼­´Â µ¥ÀÌÅÍ ¿µ¿ªÀ» °ÝÀÚµé·Î ºÐÇÒÇÏ¿© °¢ °ÝÀÚ¿¡ ¼ÓÇÑ µ¥ÀÌÅ͵éÀ» ¾î¶² ÇÑ ±â±â¿¡¼­ ¸ðµÎ ¼öÁýÇÏ¿© Áö¿ª ±ºÁýÀ» »ý¼ºÇÔ¿¡ µû¶ó ¸Ê¸®µà½º¿¡ Âü¿©ÇÏ´Â ¸ðµç ±â±âµéÀ» ¿ÏÀüÈ÷ »ç¿ëÇÑ´Ù. ½ÇÇè¿¡¼­´Â Á¦¾È ±â¹ýÀÇ ¼º´ÉÀ» ´Ù¹æ¸é¿¡ °ÉÃÄ Æò°¡Çϱâ À§ÇØ, ¿©·¯ Å©±âÀÇ ½ÇÇè µ¥ÀÌÅÍ ÁýÇÕµéÀ» »ý¼ºÇÏ°í ´Ù¾çÇÑ ½ÇÇè ȯ°æÀ» ±¸¼ºÇÏ¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
An outlier means that a data is distant from other data or a data has a contradictory relation with other data in a data set. As these features, since the outlier is used to an unnecessary data or a significant event, the outlier detection is very important. However, since the advent of the devices that capable of communication with others and environments measurement has come an explosion of data, the existing outlier detection algorithms are not suitable to the massive data. Thus, in this paper, we propose an efficient outlier detection algorithms based on data clustering over the massive data. In proposed algorithms, we utilize the MapReduce framework which comprises a number of machines in order to construct the local clusters in parallel and merge the local clusters to the global clusters. To fully utilize every machine in MapReduce, we divide the domain of the data set into a number of grids, and then each machine collects every data in each grid and constructs the local clusters. In our experiments, we created the various size of the data sets and form the diverse experiment environments in order to evaluate proposed algorithms in various fields.
Å°¿öµå(Keyword) ÀÌ»óÄ¡   µ¥ÀÌÅÍ ±ºÁýÈ­   ´ë¿ë·® µ¥ÀÌÅÍ   ¸Ê¸®µà½º   outlier   data clustering   massive data   MapReduce  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå