Probabilistic Memory Based Collaborative Filtering: Learning Individual and Social Preferences

IEEE Transactions on Knowledge and Data Engineering, 2004.

Cited by: 0|Bibtex|Views41|
Keywords:
profile density modelgraphical modeldata samplingneural networkresearch interestMore(11+)
Weibo:
In this paper we proposed a probabilistic framework for memory-based collaborative filtering

Abstract:

Memory-based collaborative filtering (CF) has been studied extensively in the literature and has proven to be successful in various types of personalized recommender systems. In this paper we develop a probabilistic framework for memory-based CF (PMCF). While this framework has clear links with classical memory-based CF, it allows us to f...More

Code:

Data:

0
Introduction
  • Information on the web has been growing explosively in recent years. Information filters emerged to meet the challenge of information search on the WWW, a problem which may be compared to “locating needles in a haystack that is growing exponentially” [1].
  • Recommender systems on e-commerce web sites assist users to find their favorite CDs or books.
  • Collaborative filtering does not rely on the content descriptions of items, but purely depends on preferences expressed by a set of users.
  • These preferences can either be expressed explicitly by numeric ratings, or can be
Highlights
  • Information on the web has been growing explosively in recent years
  • Content-based filtering (CBF) and collaborative filtering (CF) are two technologies used in recommender systems
  • We introduce probabilistic memory-based collaborative filtering (PMCF), a probabilistic framework for collaborative filtering systems that is similar in spirit to the classical memory-based collaborative filtering approach
  • In this paper we proposed a probabilistic framework for memory-based collaborative filtering (PMCF)
  • The probabilistic memory-based collaborative filtering is based on user profiles in a specially constructed profile space
  • Our results showed that the active learning approach performed better than other methods for learning user profiles, in the sense that it can make accurate predictions with only a minimum amount of user input
Methods
  • Several of the approaches pro-

    (13) posed in the active learning literature may be adopted for CF.
  • The variance of predictions var p is an appropriate measure of uncertainty.
  • An advantage of this approach lies in its low computational cost, since the authors only have to compute the predictions p for all yet unrated items
Results
  • 5) Terminate, if the profile space has reached a given maximum size or if the reduction of KL-divergence is below a given threshold.
  • It has been suggested in [30] that subsets of size |C| = 59 can be guaranteed to select profiles that are better than 95% of all other profiles with confidence 95%.
  • Precision and recall of PMCF are typically about 2-3% better than those of the competing methods
Conclusion
  • In this paper the authors proposed a probabilistic framework for memory-based collaborative filtering (PMCF).
  • The PMCF is based on user profiles in a specially constructed profile space.
  • With PMCF the posterior distribution of user ratings can be used to predict an active user’s ratings.
  • The authors showed in Sec. III how an active learning approach can be integrated smoothly into the PMCF framework.
  • The authors' results showed that the active learning approach performed better than other methods for learning user profiles, in the sense that it can make accurate predictions with only a minimum amount of user input
Summary
  • Introduction:

    Information on the web has been growing explosively in recent years. Information filters emerged to meet the challenge of information search on the WWW, a problem which may be compared to “locating needles in a haystack that is growing exponentially” [1].
  • Recommender systems on e-commerce web sites assist users to find their favorite CDs or books.
  • Collaborative filtering does not rely on the content descriptions of items, but purely depends on preferences expressed by a set of users.
  • These preferences can either be expressed explicitly by numeric ratings, or can be
  • Methods:

    Several of the approaches pro-

    (13) posed in the active learning literature may be adopted for CF.
  • The variance of predictions var p is an appropriate measure of uncertainty.
  • An advantage of this approach lies in its low computational cost, since the authors only have to compute the predictions p for all yet unrated items
  • Results:

    5) Terminate, if the profile space has reached a given maximum size or if the reduction of KL-divergence is below a given threshold.
  • It has been suggested in [30] that subsets of size |C| = 59 can be guaranteed to select profiles that are better than 95% of all other profiles with confidence 95%.
  • Precision and recall of PMCF are typically about 2-3% better than those of the competing methods
  • Conclusion:

    In this paper the authors proposed a probabilistic framework for memory-based collaborative filtering (PMCF).
  • The PMCF is based on user profiles in a specially constructed profile space.
  • With PMCF the posterior distribution of user ratings can be used to predict an active user’s ratings.
  • The authors showed in Sec. III how an active learning approach can be integrated smoothly into the PMCF framework.
  • The authors' results showed that the active learning approach performed better than other methods for learning user profiles, in the sense that it can make accurate predictions with only a minimum amount of user input
Tables
  • Table1: ACCURACY OF PREDICTIONS, MEASURED BY MEAN ABSOLUTE ERROR MAE, OF DIFFERENT CF METHODS. DETAILS ON THE INDIVIDUAL EXPERIMENTS ARE GIVEN IN SEC. V-B AND V-C. BOTH PMCF P AND PMCF D CONSISTENTLY OUTPERFORM THE COMPETING METHOD, IN PARTICULAR WHEN LITTLE INFORMATION IS GIVEN ABOUT THE ACTIVE USER IN THE GIVEN5 SCENARIO. THE RESULTS SHOWN HERE ARE BASED ON THE TRAINING/TEST SPLIT REPORTED IN SEC. V-B.3. ADDITIONAL EXPERIMENTS WITH 5 RANDOM SPLITS AND PAIRED t-TEST CONFIRMED THAT
  • Table2: ACCURACY OF RECOMMENDATIONS, MEASURED BY PRECISION AND RECALL, OF DIFFERENT CF METHODS. ALL RESULTS IN THIS
Download tables as Excel
Reference
  • D. Billsus and M. J. Pazzani, “Learning collaborative information filters,” in Proceedings of the 15th International Conference on Machine Learning. 1998, pp. 46–54, Morgan Kaufmann, San Francisco, CA.
    Google ScholarLocate open access versionFindings
  • M. Balabanovic and Y. Shoham, “Fab: Content-based, collaborative recommendation,” Communications of the ACM, vol. 40, no. 3, pp. 66–72, 1997.
    Google ScholarLocate open access versionFindings
  • R.J. Mooney and L. Roy, “Content-based book recommending using learning for text categorization,” in Proceedings of the Fifth ACM Conference on Digital Libaries, San Antonio, US, 2000, pp. 195–204, ACM Press, New York, US.
    Google ScholarLocate open access versionFindings
  • M. Pazzani, J. Muramastsu, and D. Billsus, “Syskill and webert: Identifying interesting web sites,” in Proceedings of the Thirteenth National Conference on Artificial Intelligence, Portland, OR, August 1996, pp. 54– 61.
    Google ScholarLocate open access versionFindings
  • P. Resnick, N. Iacovou, M. Sushak, P. Bergstrom, and J. Riedl, “Grouplens: An open architecture for collaborative filtering of netnews,” in Proceedings of the 1994 Computer Supported Collaborative Work Conference, Chapel Hill, North Carolina, 1994, pp. 175–186, ACM.
    Google ScholarLocate open access versionFindings
  • U. Shardanand and P. Maes, “Social information filtering algorithms for automating ’word of mouth’,” in Proceedings of ACM CHI’95 Conference on Human Factors in Computing Systems, 1995, vol. 1, pp. 210–217.
    Google ScholarLocate open access versionFindings
  • B. Sarwar, G. Karypis, J. Konstan, and J. Riedl, “Analysis of recommendation algorithms for e-commerce,” in Proceedings ACM E-Commerce Conference, 2000, pp. 158–167.
    Google ScholarLocate open access versionFindings
  • W. Hill, L. Stead, M. Rosenstein, and G. Furnas, “Recommending and evaluating choices in a virtual community of use,” in Proceedings of ACM CHI’95 Conference on Human Factors in Computing Systems, 1995, pp. 194–201.
    Google ScholarLocate open access versionFindings
  • B.J. Dahlen, J.A. Konstan, J.L. Herlocker, and J. Riedl, “Jump-starting movielens: User benefits of starting a collaborative filtering system with dead data,” Tech. Rep. 7, University of Minnesota, 1998.
    Google ScholarFindings
  • K. Goldberg, T. Roeder, D. Gupta, and C. Perkins, “Eigentaste: A constant time collaborative filtering algorithm,” Information Retrieval Journal, vol. 4, no. 2, pp. 133–151, 2001.
    Google ScholarLocate open access versionFindings
  • J. S. Breese, D. Heckerman, and C. Kadie, “Empirical analysis of predictive algorithms for collaborative filtering,” in Proceedings of the 14th Conference on Uncertainty in Artificial Intelligence, 1998, pp. 43–52.
    Google ScholarLocate open access versionFindings
  • Chumki Basu, Haym Hirsh, and William W. Cohen, “Recommendation as classification: Using social and content-based information in recommendation,” in Proceedings of the Fifteenth National Conference on Artificial Intelligencen AAAI/IAAI, 1998, pp. 714–720.
    Google ScholarLocate open access versionFindings
  • T. Zhang and V. S. Iyengar, “Recommender systems using linear classifiers,” Journal of Machine Learning Research, vol. 2, pp. 313–334, 2002.
    Google ScholarLocate open access versionFindings
  • D. Heckerman, D.M. Chickering, C. Meek, R. Rounthwaite, and C. Kadie, “Dependency networks for inference, collaborative filtering, and data visualization,” Journal of Machine Learning Research, vol. 1, pp. 49–75, 2000.
    Google ScholarLocate open access versionFindings
  • T. Hofmann and J. Puzicha, “Latent class models for collaborative filtering,” in Proceedings of IJCAI’99, 1999, pp. 688–693.
    Google ScholarLocate open access versionFindings
  • W.S. Lee, “Collaborative learning for recommender systems,” in Proc. 18th International Conf. on Machine Learning, 2001, pp. 314–321.
    Google ScholarLocate open access versionFindings
  • B. Sarwar, G. Karypis, J. Konstan, and J. Riedl, “Item-based collaborative filtering recommendation algorithms,” in Proceedings of 10th World Wide Web (WWW10) conference, Hong Kong, 2001, pp. 285–295.
    Google ScholarLocate open access versionFindings
  • W. Lin, S.A. Alvarez, and C. Ruiz, “Collaborative recommendation via adaptive association rule mining,” Data Mining and Knowledge Discovery, vol. 6, no. 1, pp. 83–105, Jan 2002.
    Google ScholarLocate open access versionFindings
  • D. M. Pennock, E. Horvitz, S. Lawrence, and C.L. Giles, “Collaborative filtering by personality diagnosis: A hybrid memory- and model-based approach,” in Proc. of the 16th Conference on Uncertainty in Artificial Intelligence, 2000, pp. 473–480.
    Google ScholarLocate open access versionFindings
  • N. Good, J.B. Schafer, J. Konstan, A. Borchers, B. Sarwar, J. Herlocker, and J. Riedl, “Combining collaborative filtering with personal agents for better recommendations,” in Proceedings of AAAI-99, 1999, pp. 439–446.
    Google ScholarLocate open access versionFindings
  • A.M. Rashid, I. Albert, D. Cosley, S.K. Lam, S.M. McNee, J.A. Konstan, and J. Riedl, “Getting to know you: Learning new user preferences in recommender systems,” in Proceedings of International Conference on Intelligent User Interface (IUI2002), San Fransisco, CA, 2002.
    Google ScholarLocate open access versionFindings
  • J.L. Herlocker, J.A. Konstan, and J. Riedl, “Explaining collaborative filtering recommendations,” in Proceedings of computer supported cooperative work (CSCW’00) conference, 2000, pp. 241–250.
    Google ScholarLocate open access versionFindings
  • Craig Boutilier and Richard S. Zemel, “Online queries for collaborative filtering,” in Ninth International Workshop on Artificial Intelligence and Statistics, 2003.
    Google ScholarLocate open access versionFindings
  • Finn V. Jensen, Bayesian Networks and Decision Graphs, Statistics for Engineering and Information Science. Springer, 2001.
    Google ScholarFindings
  • D. Heckerman, J. Breese,, and K. Rommelse, “Troubleshooting under uncertainty,” Tech. Rep. MSR-TR-94-07, Microsoft Research, 1994.
    Google ScholarLocate open access versionFindings
  • Simon Tong, Active Learning: Theory and Applicaitons, Ph.D. thesis, Stanford University, 2001.
    Google ScholarFindings
  • D. Lewis and J. Catlett, “Heterogeneous uncertainty sampling for supervised learning,” in Proceedings of the Eleventh International Conference on Machine Learning. 1994, pp. 148–156, Morgan Kaufmann.
    Google ScholarLocate open access versionFindings
  • Thomas Cover and Joy Thomas, Elements of Information Theory, Wiley, 1991.
    Google ScholarFindings
  • George Fishman, Monte Carlo Concepts, Algorithms and Applications, Springer Verlag, 1996.
    Google ScholarFindings
  • Bernhard Scholkopf and Alex J. Smola, Learning with Kernels, MIT Press, 2002.
    Google ScholarFindings
  • J.L. Herlocker, J.A. Konstan, A. Borchers, and J. Riedl, “An algorithmic framework for performing collaborative filtering,” in Proceedings of the ACM Conference on Research and Development in Information Retrieval (SIGIR’99). 1999, pp. 230–237, ACM.
    Google ScholarLocate open access versionFindings
  • Koji Miyahara and Michael J. Pazzani, “Collaborative filtering with the simple bayesian classifier,” in Proceedings of the Sixth Pacific Rim International Conference on Artificial Intelligence PRICAI 2000, 2000, pp. 679–689.
    Google ScholarLocate open access versionFindings
  • Kai Yu, Anton Schwaighofer, Volker Tresp, Wei-Ying Ma, and HongJiang Zhang, “Collaborative ensemble learning: Combining collaborative and content-based information filtering,” Submitted to UAI 2003, Apr. 2003. Kai Yu is a PhD student in the Institute for Computer Science at the University of Munich. His research work is supported through a scholarship from Siemens Corporate Technology in Munich. He has been working in speech separation, noise reduction, information retrieval and data mining. Currently his research interests are mainly focused on statistical machine learning and its applications in data mining, information and image retrieval, and medical data analysis. He received a BS and an MS in 1998 and 2000, respectively, from Nanjing University, China.
    Google ScholarLocate open access versionFindings
  • Anton Schwaighofer received his MSc degree in computer science from Graz University of Technology, Austria, in 2000. He is currently a PhD student at Graz University of Technology in co-operation with Siemens Corporate Technology in Munich. He has been working in pattern recognition for biometric applications and medical diagnosis systems. His major research interests are kernel based learning systems, in particular Gaussian processes for large scale regression problems, graphical models and clustering methods.
    Google ScholarFindings
  • Volker Tresp received a Diploma degree in physics from the University of Gottingen, Germany, in 1984 and the MSc and PhD degrees from Yale University, New Haven, CT, in 1986 and 1989, respectively. He joined the central research and development unit of Siemens AG in 1989 where he currently is the head of a research team. In 1994 he was a visiting scientist at the Massachusetts Institute of Technology’s Center for Biological and Computational Learning. His main interests include learning systems, in particular neural networks and graphical models and medical decision support systems. He has published papers on various topics including the combination of rulebased knowledge and neural networks, the problem of missing data in neural networks, time-series modeling with missing and noisy data, committee machines, learning structure in graphical models, and kernel based learning systems. He is co-editor of Neural Information Processing Systems, 13.
    Google ScholarLocate open access versionFindings
  • Xiaowei Xu joined the Department of Information Science, University of Arkansas at Little Rock as an Associate Professor, from Corporate Technology, Siemens AG, Munich, Germany in 2002. His specialty is in data mining and database management systems. His recent research interests focus on text mining, database systems for biological and medical applications, and multimodal information retrieval. Dr. Xu has been active member of ACM. He obtained his PhD from the University of Munich in 1998.
    Google ScholarFindings
  • Hans-Peter Kriegel is a full professor for database systems in the Institute for Computer Science at the University of Munich. His research interests are in spatial and multimedia database systems, particularly in query processing, performance issues, similarity search, dimensional indexing, and in parallel systems. Data exploration using visualization led him to the area of knowledge discovery and data mining. Kriegel received his MS and PhD in 1973 and 1976, respectively, from the University of Karlsruhe, Germany. Hans-Peter Kriegel has been chairman and program committee member in many international database conferences. He has published over 200 refereed conference and journal papers. In 1997 HansPeter Kriegel received the internationally prestigious “SIGMOD Best Paper Award 1997” for the publication and prototype implementation “Fast Parallel Similarity Search in Multimedia Databases” together with four members of his research team.
    Google ScholarLocate open access versionFindings
Your rating :
0

 

Tags
Comments