TA3-4 31st Fuzzy System Symposium (Chofu, September 2-4, 2015) Interactive Recommendation System LeonardoKen Orihara, 1 Tomonori Hashiyama, 1

Size: px

Start display at page:

Download "TA3-4 31st Fuzzy System Symposium (Chofu, September 2-4, 2015) Interactive Recommendation System LeonardoKen Orihara, 1 Tomonori Hashiyama, 1"

しょうこみねむら
4 years ago
Views:

1 Interactive Recommendation System LeonardoKen Orihara, 1 Tomonori Hashiyama, 1 Shun ichi Tano 1 1 Graduate School of Information Systems, The University of Electro-Communications Abstract: The performance of recommendation systems is mainly measured by the accuracy how their results meet to the users preference. User satisfaction is another index to measure their performance. In this paper, we will try to improve the users feeling of self-determination to the recommendation system by using 2 approaches. In the first approach, we explain the reason why the information was recommended. In the second approach, we show visually the position of the recommended information by mapping on the evaluation axis determined apriori. In these approach, the user can get a lot of self-determination feeling and we will see the system improvement in raising the users satisfaction. 1 Web Web [Herlocker 04]McNee [McNee 06] Yan [Yan 14] Deci[Deci 80] 1 [Deci 80]

2 [Herlocker 04] [Hijikata 12] Sinha [Shinha 02] [ 13] Youtube Web 1 A (1) A = {a 1, a 2,..., a i,..., a N } (1) (1) a i i N i a i (2) a i = (w i,1, w i,2,..., w i,j,..., w i,v ) (2) (2) w i,j i j tf idf V (3) 2.2 u = A c i a i (3) [Deci 80] Yan [Yan 14] (3) c i i i c i = 1 c i = 0 (4) K sim k = u d k u d k (4) (4) d k D k (5) R R = {r 1, r 2,..., r k,..., r K } (5) R (6) (6) s k = u d k (6) 355

3 (7) k t k 3.2 t k = arg max(s k ) (7) k 2 A (8) A = {a 1, a 2,..., a i,..., a M } (8) (8) a i i M i a i (2) A (9) e e = A a i (9) (10) sim k = e r k e r k (10) (10) r k R k R (5) 3.1 (4) r k (11) sim user = u e u e (11) l 1: 4 高い ( Y 軸 )!"#$'& 4.1 %#()*+,-1/0 '#()*+,-./& %#()*+,-1/& '#()*+,-1/& 低い ( Y 軸 ) 低い ( X 軸 ) 23456& 789:;<&!"#$%& %#()*+,-./& '#()*+,-./& %#()*+,-./0 '#()*+,-1/& 高い ( X 軸 ) 2: 2 4 Web 50 3 (3) (4) (7) K = Net Promoter Score NPS [Reichheld 06] 356

4.1 0 10 11 3: NPS Fred Reichheld NPS 0 10 11 0 10 10

4 : NPS Fred Reichheld NPS NPS (12) NPS -100% 100% NP S = (12) (4) : NPS 1 5 NPS NPS % NPS % -8.73% NPS NPS

1.0 0.59 NPS 1 5 3.01 (7) arg max 6 t k 5 1 2 NPS 5.2 NPS 2 7 NPS -83.09 3.84 7 1.0 0.68 5.3 2 NPS 1.0 0.07 30 NPS 50 1: 4 NPS[%] 1-86.

5 NPS (7) arg max 6 t k NPS 5.2 NPS 2 7 NPS NPS NPS 50 1: 4 NPS[%] : 6: t k

2: 3 NPS[%] 1 2 5 NPS -82.61-66.67-100 -83.09 [Deci 80] Deci, E, L.: The psychology of selfdetermination, D. C. Heath and company, pp. 5-53, 1980 [Herlocker 04] Herlocker, J, L., Konstan, J, A.

6 2: 3 NPS[%] NPS [Deci 80] Deci, E, L.: The psychology of selfdetermination, D. C. Heath and company, pp. 5-53, 1980 [Herlocker 04] Herlocker, J, L., Konstan, J, A., Terveen, L, G., and Riedl, J, T.: Evaluating collaborative filtering recommender systems, ACM Transactions of Information Systems, pp. 5-53, : 1 Net Promoter Score NPS NPS -8.73% NPS [Hijikata 12] Y, Hijikata., Y, Kai., and S. Nishida.: The relation between user intervention and user satisfaction for information recommendation, SAC 12 Proceedings of the 27th Annual ACM Symposium on Applied Computing, pp , 2012 [McNee 06] McNee, S. M. and Riedl, J. and Konstan, J.A.: Accurate is not always good: How Accuracy Metrics have hurt Recommender Systems, Extended Abstracts of the 2006 ACM Conference on Human Factors in Computing Systems (CHI 2006), pp , 2006 [Reichheld 06] Reichheld, F.: The Ultimate Question: Driving Good Profits and True Growth, Harvard Business School Pr, 2006 [Shinha 02] Sinha, R., and Swearingen, K.: The role of transparency in recommender systems., CHI 02 Extended Abstracts on Human Factors in Computing Systems (CHI 2002), pp , 2002 [Yan 14] Yan Lin.:, 78, 2014 [ 13],,.:,, Vol. 25, No. 1, pp , leonardo@media.is.uec.ac.jp 359

2 3, 4, 5 6 2. [1] [2] [3]., [4], () [3], [5]. Mel Frequency Cepstral Coefficients (MFCC) [9] Logan [4] MFCC MFCC Flexer [10] Bogdanov2010 [3] [14],,,

2 3, 4, 5 6 2. [1] [2] [3]., [4], () [3], [5]. Mel Frequency Cepstral Coefficients (MFCC) [9] Logan [4] MFCC MFCC Flexer [10] Bogdanov2010 [3] [14],,, DEIM Forum 2016 E1-4 525-8577 1 1-1 E-mail: is0111rs@ed.ritsumei.ac.jp, oku@fc.ritsumei.ac.jp, kawagoe@is.ritsumei.ac.jp 373 1.,, itunes Store 1, Web,., 4,300., [1], [2] [3],,, [4], ( ) [3], [5].,,.,,,,