Item-based top-N recommendation algorithms

By Mukund Deshpande, George Karypis. 2004

2,468 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Information Systems, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Information Systems

In Signal Processing

Citations in the corpus

On the Effectiveness of Linear Models for One-Class Collaborative Filtering (2016, methodology): “…the similarity matrix S ∗ as a prede-ﬁned function of R , such as cosine similarity (Sarwar et al. 2001; Linden, Smith, and York 2003): for item-based meth-ods, this is given by Other examples are the Pearson correlation, Jaccard coef-ﬁcient and conditional probability (Deshpande and Karypis 2004).”
BPR: Bayesian Personalized Ranking from Implicit Feedback (2009, methodology): “Our BPR optimizations for matrix factorization BPR-MF and k-nearest neighbor BPR-kNN are compared against weighted regularized matrix factorization (WR-MF) [5, 10], singular value decomposition (SVD-MF), k-nearest neighbor (Cosine-kNN) [2] and the most-popular model.”
Towards Making Systems Forget with Machine Unlearning (2015, methodology): “In particular, enabled by our approach, we created the first efficient unlearning algorithm for normalized cosine similarity [37, 63] commonly used by recommendation systems (e.g., LensKit) and for one-class support vector machine (SVM) [71] commonly used by classification/anomaly detection systems…”
Efficient top-n recommendation for very large scale binary rated datasets (2013, methodology): “Here, we give a formal justiﬁcation of the rescaling factor they used by deﬁning the similarity as an asymmetric product of conditional probabilities.”
Improving recommendation lists through topic diversification (2005, background): “Analyses of topic diversification’s implications on user-based [11, 22] and item-based [26, 5] collaborative filtering are provided.”
Data Mining Methods for Recommender Systems (2015, background): “Deshpande and Karypis [25], for instance, propose the use of the hit rate and the”
RecBole: Towards a Unified, Comprehensive and Efficient Framework for Recommendation Algorithms (2020, methodology): “as the learning strategy. For code reuse, we implement several commonly used 8 Table 6: Implemented models in our library RecBole. Category Model Reference General recommendation Popularity - ItemKNN Deshpande and Karypis (2004) BPR Rendle et al. (2009) NeuMF He et al. (2017b) DMF Xue et al. (2017) ”
Collaborative filtering recommender systems (2013, background): “(11) as: According to the Deshpande and Karypis (2004), one of the shortcomings of an asymmetric metric is that each item tends to have high conditional probabilities with regard to the most favored items.”
A probabilistic model to resolve diversity–accuracy challenge of recommendation systems (2015, methodology): “Such models have many applications in analyzing users’ behavior [7,19,21,26].”
Evaluating Recommendation Systems (2011, background): “A measure closely related to R-score and NDCG is Average Reciprocal Hit Rank (ARHR) [14] which is an un-normalized measure that assigns a utility 1/k to a successful recommendation at position k.”
Recommender Systems (2016, neutral): “The lift index is discussed in [361], whereas the average reciprocal hit rate (ARHR) is proposed in [181].”
Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions (2005, background): “In addition, [29], [91] present empirical evidence that item-based algorithms can provide better computational performance than traditional user-based collaborative methods while, at the same time, providing comparable or better quality than the best available user-based algorithms.”
A Survey of Collaborative Filtering Techniques (2009, background): “Another similarity measure is conditional probability-based similarity [62, 63].”
Performance of recommender algorithms on top-n recommendation tasks (2010, result): “We should note that similar non-normalized neighborhood rules were mentioned by others [7, 10].”