Á¤º¸°úÇÐȸ ³í¹®Áö B : ¼ÒÇÁÆ®¿þ¾î ¹× ÀÀ¿ë
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
´ë¿ë·®ÀÇ GPS µ¥ÀÌÅÍ ºÐ¼®À» À§ÇÑ »ç¿ëÀÚ ¼±È£ °æ·Î ¸ðµ¨ Ç¥Çö |
¿µ¹®Á¦¸ñ(English Title) |
Representation of User Preferred Route Model for Large-scale GPS Data Analysis |
ÀúÀÚ(Author) |
ÃÖÁ¤È
ÀÌÇâÁø
¹Ú¿µÅÃ
Jung-Hwa Choi
Hyang-Jin Lee
Young-Tack Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 39 NO. 04 PP. 0315 ~ 0327 (2012. 04) |
Çѱ۳»¿ë (Korean Abstract) |
½º¸¶Æ®Æù¿¡ ³»ÀåµÈ ¼¾¼¸¦ ÅëÇØ ¼öÁýÇÑ ¹æ´ëÇÑ ¾çÀÇ »ç¿ëÀÚ À§Ä¡ µ¥ÀÌÅ͸¦ °ü¸®Çϱâ À§Çؼ´Â GPS µ¥ÀÌÅͷκÎÅÍ »ç¿ëÀÚÀÇ À̵¿ ÆÐÅÏÀ» ºÐ¼®ÇÏ´Â ¿¬±¸°¡ ÇÊ¿äÇÏ´Ù. º» ¿¬±¸´Â ½º¸¶Æ®Æù »ç¿ëÀÚÀÇ ´ë¿ë·® GPS¿Í ¾Æ¿ï·¯ WiFi µ¥ÀÌÅÍÀÇ ½Ã°£ °ü°è¿Í ±×µé °£ÀÇ À§Ä¡ °ü·Ã¼ºÀ» °í·ÁÇÏ¿© ÇϳªÀÇ ¼øÂ÷ ¸ðµ¨·Î Ç¥ÇöÇÏ´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ÀÌ ¼øÂ÷ ¸ðµ¨ÀÌ »ç¿ëÀÚÀÇ ¼±È£ °æ·Î ¸ðµ¨ÀÌ µÈ´Ù. Á¦¾ÈÇÏ´Â ¹æ¹ýÀº ´Ù¼¸ °¡Áö ÇÙ½É ±â¼úÀ» Æ÷ÇÔÇÑ´Ù. ù°, »ç¿ëÀÚ À§Ä¡ µ¥ÀÌÅ͸¦ ȹµæÇÏ¿© GPS µ¥ÀÌÅÍ ¸ðµ¨À» »ý¼ºÇÑ´Ù. µÑ°, WiFi ¸Ê¿¡ ÀÇÇÑ GPS Ŭ·¯½ºÅ͸µ ¾Ë°í¸®ÁòÀ» ÀÌ¿ëÇÏ¿© ¿ÀÂ÷¸¦ °¡Áö´Â GPS¸¦ ½ÇÁ¦ »ç¿ëÀÚ°¡ Á¸ÀçÇÑ À§Ä¡·Î º¸Á¤ÇÑ´Ù. ´ÙÀ½À¸·Î POI(point of interest) ¹ß°ß ¾Ë°í¸®ÁòÀ» ÀÌ¿ëÇÏ¿© »ç¿ëÀÚÀÇ °ü½ÉÁöÁ¡À» ã´Â´Ù. ³Ý°, ¹æÇ⼺À» °í·ÁÇÑ À¯»ç °æ·Î Ŭ·¯½ºÅ͸µ ¾Ë°í¸®ÁòÀ» ÀÌ¿ëÇÏ¿© ÀÌƲ ÀÌ»óÀÇ ±ËÀû¿¡¼ ÃßÃâµÈ ¼øÂ÷Àû POIµéÀ» ÇϳªÀÇ °æ·Î·Î º´ÇÕÇÑ´Ù. ¸¶Áö¸·À¸·Î °³³ä ±×·¡ÇÁ¸¦ »ç¿ëÇÏ¿© º´ÇÕµÈ ÇϳªÀÇ °æ·Î¸¦ ¼±È£°æ·Î ¸ðµ¨·Î Ç¥ÇöÇÑ´Ù. Á¦¾ÈÇÑ ¹æ¹ýÀ¸·Î Çѱ¹ÀÇ ÇÑ Ä·ÆÛ½º¿¡¼ ÇÑ »ç¿ëÀÚ°¡ µÎ ´Þ µ¿¾È À̵¿ÇÑ ½ÇÁ¦ GPS µ¥ÀÌÅ͸¦ ¼öÁýÇÏ¿© ½ÇÇèÇÏ¿´´Ù. ½ÇÇè °á°ú´Â ÁÖ¿Í ¿äÀÏ ´ÜÀ§·Î Ŭ·¯½ºÅ͸µ ÇÑ °á°ú, »ç¿ëÀÚ°¡ ½ÇÁ¦ À̵¿ÇÑ ½Ã³ª¸®¿À¿Í ºñ±³ÇÏ¿© Æò±ÕÀûÀ¸·Î 98% ÀÌ»ó Á¤È®ÇÑ °á°ú¸¦ º¸¿´´Ù. Á¦¾ÈÇÑ ¹æ¹ýÀº ´ë¿ë·®ÀÇ GPS µ¥ÀÌÅͷκÎÅÍ ÇϳªÀÇ »ç¿ëÀÚ ¼±È£ °æ·Î ÆÐÅÏ ¶Ç´Â ƯÁ¤ ¿äÀÏ ¹× ½Ã°£ º° »ç¿ëÀÚ ÆÐÅÏ ºÐ¼®¿¡ È°¿ëµÉ ¼ö ÀÖ´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
There is a growing need for large-scale GPS data in smart phone for management of user¡¯s location data, particularly in terms of pattern analysis of user movement. The main contribution of this paper is a sequence model - that considers the temporal and spatial relation of the numerous GPS points with a recognized WiFi network at a point. The sequence model enables users to select their preferred route. The proposed method involves five key functions: (1) generating the GPS data model from user location data; (2) correcting the GPS error using the GPS clustering algorithm and WiFi map; (3) finding the preferred location by using the POI (point of interest) detection algorithm; (4)clustering at sequential POI that exceeds two days into a route by using the spatio-temporal similar route clustering algorithm; and (5)representing the preferred route model on the basis of the clustered route using a conceptual graph. The proposed method is experimentally validated using the actual trajectory data collected over two months, for a user at a university in Korea. The experimental results indicate an average success rate of around 98% on the basis of daily and weekly comparison between an actual scenario and clustered routes. Furthermore, the proposed method is useful for pattern analysis of the preferred route on a particular day or at a particular time, against large-scale real GPS location feeds.
|
Å°¿öµå(Keyword) |
½º¸¶Æ®Æù
´ë¿ë·® GPS
¼±È£ °æ·Î ¸ðµ¨
°ü½É ÁöÁ¡
À¯»ç °æ·Î Ŭ·¯½ºÅ͸µ
Smart Phone
Large-scale GPS
Preferred-route model
POI(Point of Interest)
Similar route clustering
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|