Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)
ÇѱÛÁ¦¸ñ(Korean Title) |
À帣 »ó°ü°ü°è¸¦ »ç¿ëÇÑ ¿µÈÃßõ ¾Ë°í¸®Áò |
¿µ¹®Á¦¸ñ(English Title) |
Movie Recommendation Algorithm using Genre Correlation |
ÀúÀÚ(Author) |
ȲűÔ
±è¼º±Ç
Tae-Gyu Hwang
Sung Kwon Kim
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 26 NO. 09 PP. 0429 ~ 0434 (2020. 09) |
Çѱ۳»¿ë (Korean Abstract) |
¿µÈÀ帣´Â ÁÖÁ¦, ÁٰŸ®, ºÐÀ§±â µîÀÌ ¿ä¾àµÈ Ư¼ºÀÌ°í, °°Àº À帣ÀÇ ¿µÈµéÀº ºñ½ÁÇÑ Æ¯¼ºÀ» °¡Áö¸ç, ¿µÈ Á¦ÀÛÀÚ³ª Àü¹®°¡¿¡ ÀÇÇØ ¿µÈÀ帣°¡ ºÐ·ùµÈ´Ù. »ç¿ëÀÚ°¡ »õ·Î¿î ¿µÈ¸¦ ¼±ÅÃÇÒ ¶§, ¿µÈÀ帣·ÎºÎÅÍ À¯ÃßµÈ ³»¿ëÀ» ¹ÙÅÁÀ¸·Î ÀÇ»ç°áÁ¤Çϱ⠶§¹®¿¡, ¿µÈÀ帣´Â ¼±È£µµ¸¦ ´ëÇ¥ÇÒ ¼ö ÀÖ´Â Áß¿äÇÑ ´Ü¼°¡ µÈ´Ù. ±âÁ¸ÀÇ ¹æ¹ýÀº ¿µÈÀ帣 ºÐ¼®À» ÅëÇØ Ãßõ Á¤È®¼ºÀ» Çâ»ó½ÃÄ×Áö¸¸, Çù¾÷ ÇÊÅ͸µ ±â¹ÝÀÇ ÆòÁ¡¿¹ÃøÀ¸·Î ÀÎÇØ º¹Àâµµ°¡ Å©°í, ¿µÈºÐ·ù¿¡ »ç¿ëµÈ ¸Å°³º¯¼ö °ª¿¡ µû¸¥ ¼º´ÉÆíÂ÷°¡ Å©±â ¶§¹®¿¡ ¸ðµ¨ÀÇ ÃÖÀûÈ°¡ ¾î·Á¿ì¸ç, ºÐ·ùµÈ ¿µÈµéÀÇ Á¶ÇÕÀ¸·Î Ãßõ¸ñ·ÏÀ» ÀÛ¼ºÇϱ⠶§¹®¿¡ ÀÎÀ§ÀûÀÎ Ãßõ°á°ú¸¦ º¸¿´´Ù. º» ³í¹®¿¡¼´Â Á¦½ÃµÈ ¹®Á¦ÇØ°áÀ» À§ÇÑ ¾Ë°í¸®ÁòÀ» Á¦¾ÈÇϸç, ½ÇÇèÀ» ÅëÇØ Á¦½ÃµÈ ¹®Á¦µéÀÌ ÇØ°áµÊÀ» È®ÀÎÇÒ ¼ö ÀÖ¾ú´Ù
|
¿µ¹®³»¿ë (English Abstract) |
Genres categorize movies and help summarize their themes, plots and moods. As such, movies of the same genre should have similar characteristics when they are classified by movie-makers or domain experts. When you choose a new movie, the genre becomes an important clue to match your preferences to something you haven¡¯t seen before, basically it helps us make decisions based on analogy to previous movies we have seen in that genre. Although previous methods have improved recommendation accuracy through analysis of movie genre, recommending movies is a high complexity problem due to the collaborative filtering-based rating prediction used. Also, this model is difficult to optimize because of the large variation in performance according to the parameter values used in the movie classification, at the same time the recommended results are artificial because the recommendation list is written using a combination of classified movies. In this paper, we introduce new methods for solving the problems presented, and through experiments we show that our approach successfully solves these problems
|
Å°¿öµå(Keyword) |
ÆíÇâ ±â¹Ý ºÐ¼®
À帣 »ó°ü°ü°è
ÆòÁ¡ ¿¹Ãø
À帣 ±â¹Ý Ãßõ
Ãßõ ¾Ë°í¸®Áò
bias-based analysis
genre correlation
rating prediction
genre-based recommendation
recommendation algorithm
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|