AI帮你理解科学

AI 生成解读视频

AI抽取解析论文重点内容自动生成视频


pub
生成解读视频

AI 溯源

AI解析本论文相关学术脉络


Master Reading Tree
生成 溯源树

AI 精读

AI抽取本论文的概要总结


微博一下
Based on the intuition that every user’s decisions on the Web should include both the user’s characteristics and the user’s trusted friends’ recommendations, we propose a novel, effective and efficient probabilistic matrix factorization framework for the recommender systems

Learning to recommend with social trust ensemble

SIGIR, pp.203-210, (2009)

被引用932|浏览267
EI
下载 PDF 全文
引用
微博一下

摘要

As an indispensable technique in the field of Information Filtering, Recommender System has been well studied and developed both in academia and in industry recently. However, most of current recommender systems suffer the following problems: (1) The large-scale and sparse data of the user-item matrix seriously affect the recommendation q...更多

代码

数据

0
简介
  • As the exponential growth of information generated on the World Wide Web, the Information Filtering techniques like Recommender Systems have become more and more important and popular.
  • Recommender systems are based on Collaborative Filtering, which is a technique that automatically predicts the interest of an active user by collecting rating information from other similar users or items.
  • Many collaborative filtering algorithms are impeded by the sparsity problem, cannot handle users who have rated few items.
  • Trust-aware recommender systems have drawn lots of attention [14, 15], but most of these methods are based on some ad hoc heuristics, and they still have the data sparsity and scalability problems.
  • The relationship between the user-item matrix and the users’ trust network are not fully understood
重点内容
  • As the exponential growth of information generated on the World Wide Web, the Information Filtering techniques like Recommender Systems have become more and more important and popular
  • (2) The traditional recommender systems assume that all the users are independent and identically distributed; this assumption ignores the connections among users, which is not consistent with the real world recommendations
  • Aiming at modeling recommender systems more accurately and realistically, we propose a novel probabilistic factor analysis framework, which naturally fuses the users’ tastes and their trusted friends’ favors together
  • Recommender systems are based on Collaborative Filtering, which is a technique that automatically predicts the interest of an active user by collecting rating information from other similar users or items
  • Based on the intuition that every user’s decisions on the Web should include both the user’s characteristics and the user’s trusted friends’ recommendations, we propose a novel, effective and efficient probabilistic matrix factorization framework for the recommender systems
  • Under the circumstance that both the user-item rating matrix and the trust relations of a social network are very sparse, the diffusions of trust relations become inevitable since this consideration will help to alleviate the data sparsity problem and will potentially increase the prediction accuracy
结果
  • As reported in [20], the density of the available ratings in commercial recommender systems is often less than 1%.
  • The density of the useritem rating matrix is less than 0.015%.
  • From Fig. 4(a) and Fig. 4(b), when using 90% ratings as training data, the authors observe that, the RSTE method achieves the best performance when α is around 0.4, while smaller values like α = 0.1 or larger values like α = 0.7 can potentially degrade the model performance
结论
  • CONCLUSIONS AND FUTURE WORK

    This paper is motivated by the fact that a user’s trusted friends on the Web will affect this user’s online behavior.
  • Based on the intuition that every user’s decisions on the Web should include both the user’s characteristics and the user’s trusted friends’ recommendations, the authors propose a novel, effective and efficient probabilistic matrix factorization framework for the recommender systems.
  • The method introduced in this paper by using probabilistic matrix factorization is working in trust-aware recommender systems, and applicable to other popular research topics, such as social search, collaborative information retrieval, and social data mining.
  • The authors plan to employ the diffusion processes in the future work
总结
  • Introduction:

    As the exponential growth of information generated on the World Wide Web, the Information Filtering techniques like Recommender Systems have become more and more important and popular.
  • Recommender systems are based on Collaborative Filtering, which is a technique that automatically predicts the interest of an active user by collecting rating information from other similar users or items.
  • Many collaborative filtering algorithms are impeded by the sparsity problem, cannot handle users who have rated few items.
  • Trust-aware recommender systems have drawn lots of attention [14, 15], but most of these methods are based on some ad hoc heuristics, and they still have the data sparsity and scalability problems.
  • The relationship between the user-item matrix and the users’ trust network are not fully understood
  • Results:

    As reported in [20], the density of the available ratings in commercial recommender systems is often less than 1%.
  • The density of the useritem rating matrix is less than 0.015%.
  • From Fig. 4(a) and Fig. 4(b), when using 90% ratings as training data, the authors observe that, the RSTE method achieves the best performance when α is around 0.4, while smaller values like α = 0.1 or larger values like α = 0.7 can potentially degrade the model performance
  • Conclusion:

    CONCLUSIONS AND FUTURE WORK

    This paper is motivated by the fact that a user’s trusted friends on the Web will affect this user’s online behavior.
  • Based on the intuition that every user’s decisions on the Web should include both the user’s characteristics and the user’s trusted friends’ recommendations, the authors propose a novel, effective and efficient probabilistic matrix factorization framework for the recommender systems.
  • The method introduced in this paper by using probabilistic matrix factorization is working in trust-aware recommender systems, and applicable to other popular research topics, such as social search, collaborative information retrieval, and social data mining.
  • The authors plan to employ the diffusion processes in the future work
表格
  • Table1: Statistics of User-Item Rating Matrix of Epinions
  • Table2: Statistics of Social Trust Network of Epinions
  • Table3: Performance Comparisons (A Smaller MAE or RMSE Value Means a Better Performance)
Download tables as Excel
相关工作
  • In this section, we review several major approaches for recommender systems, including (1) traditional recommender systems which are mainly based on collaborative filtering techniques, and (2) social trust-based recommender systems which have drawn lots of attention recently.

    Traditional collaborative filtering algorithms mainly focus on the user-item matrix. Among all of these methods, the memory-based approaches are the most popular methods and they are widely adopted in commercial collaborative filtering systems [10, 17]. These methods employ different strategies to find similar users and items for making the predictions, which are known as user-based approaches [3, 6, 9, 12] and item-based approaches [5, 10, 20], respectively. To predict a rating Rij of a given item vj for an active user ui, user-based methods search for other users similar to the user ui and utilize their ratings to the item vj for prediction, while item-based methods leverage the ratings of other items similar to the item vj from the user ui instead. In order to take advantages of these two types of methods, Wang et al in [23] and Ma et al in [12] proposed two fusion models to combine user-based method with itembased method. In addition to the memory-based methods, model-based approaches, which employ statistical and machine learning techniques to learn models from the data, also play an important role in the collaborative filtering research. Examples of model-based approaches include aspect models [7, 8, 21], the latent factor model [4], the Bayesian hierarchical model [24] and the ranking model [11]. Recently, several matrix factorization methods [16, 18, 19, 22] have been proposed for collaborative filtering. These methods focus on factorizing the user-item rating matrix using lowrank representations, and then utilize them to make further predictions. The motivation behind a low-dimensional factorization model is that there is only a small number of
基金
  • The work described in this paper is supported by grants from the Research Grant Council of the Hong Kong Special Administrative Region, China (Project No.: CUHK 4128/08E and CUHK 4158/08E)
研究对象与分析
users: 5
In the real world, the process of recommendation scenario includes two central elements: the trust network and the favors of these friends, which can essentially be modeled by the examples of the trust graph in Fig. 1(a) and the useritem rating matrix in Fig. 1(b), respectively. In the trust graph illustrated in Fig. 1(a), totally, 5 users (nodes, from u1 to u5) are connected with 9 relations (edges) between users, and each relation is associated with a weight Sij in the range (0, 1] to specify how much user ui knows or trusts user uj. Normally, the trust relations in the online trust network are explicitly stated by online users

users: 51670
The dataset used in our experiments is collected by crawling the Epinions.com site on Jan 2009. It consists of 51,670 users who have rated a total of 83,509 different items. The total number of ratings is 631,064

users: 6040
0.75 1−10 11−20 21−40 41−80 81−160 >160 Number of Observed Ratings (b) MAE Comparison on Different User Rating Scales (90% as Training Data). Number of Observed Ratings (c) RMSE Comparison on Different User Rating Scales (90% as Training Data) the user-item rating matrix of Epinions is very sparse, since the densities for the two most famous collaborative filtering datasets Movielens (6,040 users, 3,900 movies and 1,000,209 ratings) and Eachmovie (74,424 users, 1,648 movies and 2,811,983 ratings) are 4.25% and 2.29%, respectively. Moreover, an important factor that we choose the Epinions dataset is that user social trust network information is not included in the Movielens and Eachmovie datasets

引用论文
  • R. Andersen, C. Borgs, J. Chayes, U. Feige, A. Flaxman, A. Kalai, V. Mirrokni, and M. Tennenholtz. Trust-based recommendation systems: an axiomatic approach. In Proc. of WWW ’08, pages 199–208, New York, NY, USA, 2008. ACM.
    Google ScholarLocate open access versionFindings
  • P. Bedi, H. Kaur, and S. Marwaha. Trust based recommender system for semantic web. In Proc. of IJCAI ’07, pages 2677–2682, 2007.
    Google ScholarLocate open access versionFindings
  • J. S. Breese, D. Heckerman, and C. Kadie. Empirical analysis of predictive algorithms for collaborative filtering. In Proc. of UAI ’98, 1998.
    Google ScholarLocate open access versionFindings
  • J. Canny. Collaborative filtering with privacy via factor analysis. In Proc. of SIGIR ’02, pages 238–245, New York, NY, USA, 2002. ACM.
    Google ScholarFindings
  • M. Deshpande and G. Karypis. Item-based top-n recommendation. ACM Transactions on Information Systems, 22(1):143–177, 2004.
    Google ScholarLocate open access versionFindings
  • J. L. Herlocker, J. A. Konstan, A. Borchers, and J. Riedl. An algorithmic framework for performing collaborative filtering. In Proc. of SIGIR ’99, pages 230–237, New York, NY, USA, 1999. ACM.
    Google ScholarFindings
  • T. Hofmann. Collaborative filtering via gaussian probabilistic latent semantic analysis. In Proc. of SIGIR ’03, pages 259–266, New York, NY, USA, 2003. ACM.
    Google ScholarFindings
  • T. Hofmann. Latent semantic models for collaborative filtering. ACM Transactions on Information Systems, 22(1):89–115, 2004.
    Google ScholarLocate open access versionFindings
  • R. Jin, J. Y. Chai, and L. Si. An automatic weighting scheme for collaborative filtering. In Proc. of SIGIR ’04, pages 337–344, New York, NY, USA, 2004. ACM.
    Google ScholarFindings
  • G. Linden, B. Smith, and J. York. Amazon.com recommendations: Item-to-item collaborative filtering. IEEE Internet Computing, pages 76–80, Jan/Feb 2003.
    Google ScholarLocate open access versionFindings
  • N. N. Liu and Q. Yang. Eigenrank: a ranking-oriented approach to collaborative filtering. In Proc. of SIGIR ’08, pages 83–90, New York, NY, USA, 2008. ACM.
    Google ScholarLocate open access versionFindings
  • H. Ma, I. King, and M. R. Lyu. Effective missing data prediction for collaborative filtering. In Proc. of SIGIR ’07, pages 39–46, New York, NY, USA, 2007. ACM.
    Google ScholarFindings
  • H. Ma, H. Yang, M. R. Lyu, and I. King. SoRec: Social recommendation using probabilistic matrix factorization. In Proc. of CIKM ’08, pages 931–940, New York, NY, USA, 2008. ACM.
    Google ScholarLocate open access versionFindings
  • P. Massa and P. Avesani. Trust-aware collaborative filtering for recommender systems. In Proc. of CoopIS/DOA/ODBASE, pages 492–508, 2004.
    Google ScholarLocate open access versionFindings
  • P. Massa and P. Avesani. Trust-aware recommender systems. In Proc. of RecSys, pages 17–24, New York, NY, USA, 2007. ACM.
    Google ScholarFindings
  • J. D. M. Rennie and N. Srebro. Fast maximum margin matrix factorization for collaborative prediction. In Proc. of ICML ’05, 2005.
    Google ScholarLocate open access versionFindings
  • P. Resnick, N. Iacovou, M. Suchak, P. Bergstrom, and J. Riedl. Grouplens: An open architecture for collaborative filtering of netnews. In Proc. of CSCW ’94, 1994.
    Google ScholarLocate open access versionFindings
  • R. Salakhutdinov and A. Mnih. Bayesian probabilistic matrix factorization using markov chain monte carlo. In Proc. of ICML ’08, 2008.
    Google ScholarLocate open access versionFindings
  • R. Salakhutdinov and A. Mnih. Probabilistic matrix factorization. In Advances in Neural Information Processing Systems, volume 20, 2008.
    Google ScholarLocate open access versionFindings
  • B. Sarwar, G. Karypis, J. Konstan, and J. Reidl. Item-based collaborative filtering recommendation algorithms. In Proc. of WWW ’01, pages 285–295, New York, NY, USA, 2001. ACM.
    Google ScholarFindings
  • L. Si and R. Jin. Flexible mixture model for collaborative filtering. In Proc. of ICML ’03, 2003.
    Google ScholarLocate open access versionFindings
  • N. Srebro and T. Jaakkola. Weighted low-rank approximations. In Proc. of ICML ’03, 2003.
    Google ScholarLocate open access versionFindings
  • J. Wang, A. P. de Vries, and M. J. T. Reinders. Unifying user-based and item-based collaborative filtering approaches by similarity fusion. In Proc. of SIGIR ’06, pages 501–508, New York, NY, USA, 2006. ACM.
    Google ScholarFindings
  • Y. Zhang and J. Koren. Efficient bayesian hierarchical user modeling for recommendation system. In Proc. of SIGIR ’07, pages 47–54, New York, NY, USA, 2007. ACM.
    Google ScholarFindings
您的评分 :
0

 

标签
评论
数据免责声明
页面数据均来自互联网公开来源、合作出版商和通过AI技术自动分析结果,我们不对页面数据的有效性、准确性、正确性、可靠性、完整性和及时性做出任何承诺和保证。若有疑问,可以通过电子邮件方式联系我们:report@aminer.cn
小科