Modern Recommender System

Everyday human society are producing data in amazing quantity , the amount of information on the web is increasing at such an astonishing rate that people’s brain can’t process it. Search engine is a probable approach to get the information what you want, but if user’s requirements are vague, a personal recommendation system seems more significant. There are three main classes of recommendation system. Those are:

(1) Collaborative filtering systems:

Collaborative systems generate recommendations based on crowd-sourced input. They recommend for user according to find similar user group, analysis those similar users’ rating for certain item, to generate preference prediction for this user in special products(An example is Google PageRank, which recommends similar web pages based on a web pages’back links).Collaborative filtering algorithm can be generally classified as two types: Memory-based collaborative filtering and Model-based collaborative filtering.

1.1 Memory-based collaborative filtering:

Memory-based collaborative filtering include User-based collaborative filtering[1] which evaluate the similarity between users by different users’ ratings on the item, making recommendations based on the similarity between users and Item-based collaborative filtering[3,4] which evaluate the similarity between items by the user’s rating on different items, making recommendations based on the similarity between items.

1.2 Model-based collaborative filtering

With the quantity of users and items rapidly increasing, the problems of high dimensionality and sparsity of rating matrix become more and more serious, people introduce Model-based collaborative filtering. There are many model-based CF algorithms. These include Bayesian networks[5,6], clustering models[8], latent semantic models such as singular value decomposition, Probabilistic Matrix Factorization[9,10,11], probabilistic latent semantic analysis, multiple multiplicative factor, latent Dirichlet allocation and Markov decision process based models.

(2) Content-based filtering systems:

Content-based systems[2] generate recommendations for users based on a description of the item and a profile of the user’s preference (Pandora uses content-based filtering to make its music recommendations). It doesn’t need ratings data but keywords are used to describe the items and a user profile is built to indicate the type of item this user likes. A widely used algorithm is the Tf-idf[13] representation

(3) Hybrid recommendation systems:

Hybrid recommendation systems[12] combine both collaborative and content-based approaches. They help improve recommendations that are derived from sparse datasets. (Netflix is a prime example of a hybrid recommender)
Reference:
[1] Resnick P, Iacovou N, Suchak M, et al. GroupLens: an open architecture for collaborative filtering of netnews[C]// ACM Conference on Computer Supported Cooperative Work. ACM, 1994:175-186.(UBCF)
[2] Balabanovi, Marko, Shoham Y. Fab: content-based, collaborative recommendation[J]. Communications of the Acm, 1997, 40(3):66-72.(CBCF)
[3] Sarwar B, Karypis G, Konstan J, et al. Item-based collaborative filtering recommendation algorithms[C]// International Conference on World Wide Web. ACM, 2001:285-295.(IBCF)
[4] Deshpande M, Karypis G. Item-based top- N recommendation algorithms[J]. Acm Transactions on Information Systems, 2004, 22(1):143-177.(IBCF)
[5] Su X, Khoshgoftaar T M. Collaborative Filtering for Multi-class Data Using Belief Nets Algorithms[C]// IEEE International Conference on TOOLS with Artificial Intelligence. IEEE, 2006:497-504.(贝叶斯网络)
[6] Miyahara K, Pazzani M J. Collaborative Filtering with the Simple Bayesian Classifier[J]. 2002, 1886:679-689. (贝叶斯网络)
[7] Herlocker J. Clustering Items for Collaborative Filtering[J]. Proceedings of the Acm Sigir Workshop on Recommender Systems, 1999.(聚类算法)
[8] Sarwar B M, Karypis G, Konstan J, et al. Recommender systems for large-scale e-commerce: Scalable neighborhood formation using clustering[C]// Conference on Computer and Information Technology. 2002.(聚类算法)
[9] Salakhutdinov R, Mnih A. Probabilistic matrix factorization[C]// International Conference on Machine Learning. 2008:880-887.(PMF)
[10] Takács G, Pilászy I, Németh B, et al. Matrix factorization and neighbor based algorithms for the netflix prize problem[C]// ACM Conference on Recommender Systems, Recsys 2008, Lausanne, Switzerland, October. DBLP, 2008:267-274.(MF)
[11] Koren Y, Bell R, Volinsky C. Matrix factorization techniques for recommender systems. IEEE, Computer Journal, 42(8), 30-37[J]. 2009, 42(8):30-37.(MF)
[12] Burke R. Hybrid Recommender Systems: Survey and Experiments[J]. User Modeling and User-Adapted Interaction, 2002, 12(4):331-370.(Hybrid)
[13] Belkin N J, Croft W B. Information filtering and information retrieval: two sides of the same coin?[J]. Communications of the Acm, 1992, 35(12):29-38.(TF-IDF)