µ¥ÀÌÅͺ£À̽º ¿¬±¸È¸Áö(SIGDB)
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
BAR : ½ÇÃ¼ÈµÈ ºñÆ®¸Ê ±â¹ÝÀÇ ¿¬°ü ±ÔÄ¢ ¾Ë°í¸®Áò |
¿µ¹®Á¦¸ñ(English Title) |
mBAR : Materialized Bitmap based Association Rules Algorithm |
ÀúÀÚ(Author) |
±èµ¿Çö
°¿îÇÐ
ÀÌ»ó¿ø
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 20 NO. 04 PP. 0045 ~ 0062 (2004. 12) |
Çѱ۳»¿ë (Korean Abstract) |
Á¤º¸±â¼ú°ú ÀÎÅͳÝÀÇ °è¼ÓÀûÀÎ ¹ß´Þ·Î µ¥ÀÌÅÍ ¸¶ÀÌ´×Àº ´Ù¾çÇÑ ºÐ¾ß¿¡¼ È°¿ëµÇ°í ÀÖÀ¸¸ç, ±× Áß ¿¬°ü ±ÔÄ¢ ¸¶ÀÌ´×Àº ÁÖ¾îÁø Æ®·£Àè¼Ç µ¥ÀÌÅÍ ÁýÇÕ¿¡¼ °øÅëÀûÀ¸·Î ÀÚÁÖ ¹ß»ýÇÏ´Â Ç׸ñµî¸¦ ã¾Æ³»´Â ±â¹ýÀÌ´Ù. ´ëÇ¥ÀûÀÎ ¿¬°ü±ÔÄ¢ ¾Ë°í¸®ÁòÀ¸·Î´Â Apriori°¡ Àִµ¥, Áö³ 10³â µ¿¾È Áö¼ÓÀûÀÎ ¿¬±¸¸¦ ÅëÇØ ¸¹Àº ¼º´É°³¼±ÀÌ ÀÌ·ç¾îÁ³´Ù. ÇÏÁö¸¸, ±âº»ÀûÀ¸·Î ºó¹ßÇ׸ñÁýÇÕÀ» ±¸ÇÒ ¶§ ¸Å¹ø ¸ðµç Ç׸ñÁýÇյ鿡 ´ëÇØ °è»êÀ» Çϱ⠶§¹®¿¡, ¿©ÀüÈ÷ ½Ã°£ÀÌ ¸¹ÀÌ °É¸°´Ù. º» ³í¹®¿¡¼´Â ½ÇÃ¼ÈµÈ ºñÆ®¸Ê °³³äÀ» È°¿ëÇØ Apriori ¾Ë°í¸®ÁòÀ» °è»êÇÔÀ¸·Î½á ȹ±âÀûÀ¸·Î ¿¬°ü±ÔÄ¢ ¼º´ÉÀ» °³¼±ÇÏ´Â ¹æ¾ÈÀ» Á¦½ÃÇÑ´Ù. À̸¦ À§ÇØ ¿ì¼± ºñÆ®¸Ê °³³äÀ» ÀÌ¿ëÇÑ Apriori ¾Ë°í¸®Áò ¼öÇà¹æ½ÄÀ» Á¦¾ÈÇÏ°í, ¶ÇÇÑ ¸Å¹ø ºñÆ®¸ÊÀ» »õ·Î °è»êÇÏ´Â ´ë½Å¿¡ ½ÇüȽÃÅ´À¸·Î½á ºó¹ßÇ׸ñÁýÇÕÀ» ȹ±âÀûÀ¸·Î »¡¸® ã¾Æ³»´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ´õºÒ¾î¼, ÇÑÁ¤µÈ ÀúÀå °ø°£À» È°¿ëÇؼ ½ÇüȽÃų ºñÆ®¸ÊµéÀ» ¼±ÅÃÇÏ´Â ¹æ¾È°ú Æ®·£Àè¼Ç Å×ÀÌºí¿¡ º¯È°¡ »ý±æ ¶§ À̸¦ ½ÇÃ¼ÈµÈ ºñÆ®¸ÊµéÀ» Á¡ÁøÀûÀ¸·Î °ü¸®ÇÏ´Â ±â¹ýÀ» Á¦¾ÈÇÑ´Ù. |
¿µ¹®³»¿ë (English Abstract) |
With the rapid progress in information technology and the Internet, the data mining technique has been exploited in various applications. The association rule(hereafter, AR) mining, one of the most popular data mining techniques, is to find the frequent itemsets which occur commonly in transaction database. Of the various AR algorithms, the Apriori is most popular, and it has been continuously improved during the past decade. Even with the most recent version, however, it is very time consuming for the Apriori-based algorithms to count frequent itemset since, basically for each k-size item set, we need to compute its support on the fly. In this paper, we propose the mBAR approach to AR mining, which drastically improves the Apriori algorithm by exploiting materialized bitmaps. First, we present a bitmap-base Apriori algorithm. And, we suggest, in order to boost the performance of finding the frequent itemsets, how to store(i. e. materialize) the bitmaps, instead of computing the bitmaps on-the-fly. Related to the materialized bitmaps, we suggest a way to choose the bitmaps selectively, instead of full bitmaps, and propose an incremental maintenance technique for materialized bitmaps against the changes in transaction database, |
Å°¿öµå(Keyword) |
µ¥ÀÌÅÍ ¸¶ÀÌ´×
¿¬°ü±ÔÄ¢
ºñÆ®¸Ê
½ÇÃ¼ÈµÈ ºä
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|