ROC2-NewMovies (users DC2-MovieLens (users with range 1-99 ratin group 10-19 鸟一 Figure 6. Optimization, using ROC2 technique, of the SVD algorithm on: (a)New Movie dataset, users who have at most 9 ratings. (b)mL dataset, users who have at most 99 ratings. (c)shows ROC2 technique applied to the NBn algorithm on the New Movies dataset. (b) Popularity Popularity Popularity UPL First80 NoFirst80 UPL First80 NoFirst80 UPL First80 NoFirst80 2-931.6%0.13% 2-930.6%0.04% 2-929.3%0.01% 10-1924.7%0.19% 10-1926.3%0.13% 10-1923.7%0.01% >2020.7%0.01% >2021.7%0.02% >2023.2%0.02% Table 2 Recall obtained with the NBN algorithm applied on the NM dataset. Models are created with users who have, respectively, at least 2(a), 10(b)and 20 (c)ratings. UPL stands for User Profile Length [111 R. Duda, P. Hart, and D. Stork. Pattern Classification on World wide Web, pages 285-295, New York, NY, USA, Wiley-Interscience, 2000 2001. ACM Press [12] J Herlocker, J. Konstan, L. Terveen, and J. Riedl. Evaluating [17 B Sarwar, G. Karypis, J. Konstan, and J. Riedl. Analysis of recommendation algorithms for e-commerce. Proceedings [131 J. L. Herlocker, J. A. Konstan, A. Borchers, and J. Riedl. of the 2nd ACM conference on Electronic commerce, pages 158-167,2000 An algorithmic framework for performing collaborative fil- [18] B Sarwar, G. Karypis, J Konstan, J. Riedl, and M. U. M. D tering. In SIGIR 99: Proceedings of the 22nd annual in- O C SCIENCE. Application of Dimensionality Reduction in Recommender System. A Case Study. Defense Technical pment in information retrieval, pages 230-237, New York Information Center. 2000 NY USA. 1999. ACM [19] U. Shardanand and P Maes. Social information filtering [141 J. Konstan, B. Miller, D. Maltz, J. Herlocker, L. Gordon algorithms for automating"word of mouth". Proceedings and J. Riedl. GroupLens: applying collaborative filtering to of the SIGCHI conference on Human factors in computing Usenet news. Communications of the ACM, 40(3): 77-87 1997 [20]1. Witten and E. Frank. Data Mining: Practical Machine [15] M. Papagelis and D. Plexousakis. Qualitative analysis of Learning Tools and Techniques. Morgan Kaufmann, 2005. user-based and item-based prediction algorithms for recom- mendation agents. Engineering Applications of artificial telligence, 18(7): 781-789, October 2005 [16] B. Sarwar, G. Karypis, J. Konstan, and J. Reidl. Item- based collaborative filtering recommendation algorithms. In www 01: Proceedings of the 1Oth international conference0 0.005 0.01 0.015 0.02 0.025 0.03 0 0.05 0.1 0.15 0.2 0.25 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies (users with range 2!9 ratings) random curve roc curve L=15 roc curve L=50 roc curve L=100 roc curve L=200 roc curve L=300 Top!Rated (a) 0 0.002 0.004 0.006 0.008 0.01 0.012 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 False Positive Rate True Positive Rate (Recall) ROC2 ! MovieLens (users with range 1!99 ratings) random curve roc curve L=15 roc curve L=50 roc curve L=100 roc curve L=200 roc curve L=300 roc curve Top!Rated (b) 0 0.005 0.01 0.015 0.02 0.025 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies model (users with 2!9 ratings) random curve group 2!9 group 10!19 group 20!inf top!rated 0 0.005 0.01 0.015 0.02 0.025 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies model (users with 2!9 ratings) random curve group 2!9 group 10!19 group 20!inf top!rated 0 0.005 0.01 0.015 0.02 0.025 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies model (users with 2!9 ratings) random curve group 2!9 group 10!19 group 20!inf top!rated 0 0.005 0.01 0.015 0.02 0.025 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies model (users with 2!9 ratings) random curve group 2!9 group 10!19 group 20!inf top!rated 0 0.005 0.01 0.015 0.02 0.025 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 False Positive Rate True Positive Rate (Recall) ROC2 ! NewMovies model (users with 2!9 ratings) random curve group 2!9 group 10!19 group 20!inf top!rated (c) Figure 6. Optimization, using ROC2 technique, of the SVD algorithm on: (a) NewMovie dataset, users who have at most 9 ratings. (b) ML dataset, users who have at most 99 ratings. (c) shows ROC2 technique applied to the NBN algorithm on the NewMovies dataset. (a) Popularity UPL First80 NoFirst80 2-9 31.6% 0.13% 10-19 24.7% 0.19% ≥ 20 20.7% 0.01% (b) Popularity UPL First80 NoFirst80 2-9 30.6% 0.04% 10-19 26.3% 0.13% ≥ 20 21.7% 0.02% (c) Popularity UPL First80 NoFirst80 2-9 29.3% 0.01% 10-19 23.7% 0.01% ≥ 20 23.2% 0.02% Table 2. Recall obtained with the NBN algorithm applied on the NM dataset. Models are created with users who have, respectively, at least 2 (a), 10 (b) and 20 (c) ratings. UPL stands for User Profile Length. [11] R. Duda, P. Hart, and D. Stork. Pattern Classification. Wiley-Interscience, 2000. [12] J. Herlocker, J. Konstan, L. Terveen, and J. Riedl. Evaluating collaborative filtering recommender systems. ACM Transac￾tions on Information Systems (TOIS), 22(1):5–53, 2004. [13] J. L. Herlocker, J. A. Konstan, A. Borchers, and J. Riedl. An algorithmic framework for performing collaborative fil￾tering. In SIGIR ’99: Proceedings of the 22nd annual in￾ternational ACM SIGIR conference on Research and devel￾opment in information retrieval, pages 230–237, New York, NY, USA, 1999. ACM. [14] J. Konstan, B. Miller, D. Maltz, J. Herlocker, L. Gordon, and J. Riedl. GroupLens: applying collaborative filtering to Usenet news. Communications of the ACM, 40(3):77–87, 1997. [15] M. Papagelis and D. Plexousakis. Qualitative analysis of user-based and item-based prediction algorithms for recom￾mendation agents. Engineering Applications of Artificial In￾telligence, 18(7):781–789, October 2005. [16] B. Sarwar, G. Karypis, J. Konstan, and J. Reidl. Item￾based collaborative filtering recommendation algorithms. In WWW ’01: Proceedings of the 10th international conference on World Wide Web, pages 285–295, New York, NY, USA, 2001. ACM Press. [17] B. Sarwar, G. Karypis, J. Konstan, and J. Riedl. Analysis of recommendation algorithms for e-commerce. Proceedings of the 2nd ACM conference on Electronic commerce, pages 158–167, 2000. [18] B. Sarwar, G. Karypis, J. Konstan, J. Riedl, and M. U. M. D. O. C. SCIENCE. Application of Dimensionality Reduction in Recommender System. A Case Study. Defense Technical Information Center, 2000. [19] U. Shardanand and P. Maes. Social information filtering: algorithms for automating “word of mouth”. Proceedings of the SIGCHI conference on Human factors in computing systems, pages 210–217, 1995. [20] I. Witten and E. Frank. Data Mining: Practical Machine Learning Tools and Techniques. Morgan Kaufmann, 2005