Lessons from the Netflix prize challenge

By Robert M. Bell, Yehuda Koren. 2007

848 citations recorded by the source corpus.

Subfields: Information Systems, Management Information Systems, Sociology and Political Science, Recommender Systems

How researchers see this paper

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

In Information Systems

In Management Information Systems

In Sociology and Political Science

In Recommender Systems

Citations in the corpus

Some Improvements on Deep Convolutional Neural Network Based Image Classification (2013, background): “Reference s [1] R.”
How algorithmic confounding in recommendation systems increases homogeneity and decreases utility (2017, methodology): “Rating prediction is a common focus for recommendation algorithms, owing its popularity at least in part to the Netflix challenge [4, 5], which evaluated systems using RMSE on a set of held-out data.”
Ask the GRU: Multi-task Learning for Deep Text Recommendations (2016, background): “This allows us to make predictions in cold-start conditions, while providing regularization for the recommendation model.”
Recommendations : Modeling Music Ratings with Temporal Dynamics and Item Taxonomy (2011, positive): “In general, CF has proved more accurate in predicting rating datasets devoid of intrinsic item information such as the Netflix dataset [4].”
Protecting privacy when mining and sharing user data (2012, methodology): “Furthermore, the contest has given high publicity and led to significant progress in the field of recommender systems [23,107], resulting in establishment of matrix factorization techniques as the dominant methodology for implementing collaborative filtering [108, 177], introduction of two-layer und”
Quantum Recommendation Systems (2016, background): “been a series of works in the machine learning community many of them inspired by a practical challenge by Netflix on real world data [14, 6, 13].”
On the stability of recommendation algorithms (2010, background): “Prediction is done by taking inner product of user-factor and item-factor vectors [4, 6, 11].”
TopicMF: Simultaneously Exploiting Ratings and Reviews for Recommendation (2014, methodology): “Among all the CF algorithms, the most successful ones, as demonstrated by the Netflix Prize competition (Bell and Koren 2007), are the latent factor models.”
Recommended for you: The Netflix Prize and the production of algorithmic culture (2016, methodology): “Although most competitors ended up incorporating other techniques, for many, SVD formed the center-piece of their algorithms (see, for example, Koren and Bell, 2007; Koren et al., 2009).”
A Distributed, Architecture-Centric Approach to Computing Accurate Recommendations from Very Large and Sparse Datasets (2011, methodology): “However, users are usually hesitant to provide additional ratings, third party data might not be available, and implicit rating analysis is unreliable [12,102].”
Does data splitting improve prediction? (2013, background): “For example, in the Netﬂix prize competition (Bell and Koren (2007)) it was essential to hold back part of the data to evaluate the performance of the model since the competitors estimates of performance could not be trusted.”
Bayesian Hyperparameter Optimization for Ensemble Learning (2016, methodology): “We study the performance of our proposed method on three different hyperparameter spaces, showing that our approach is better than both the best single model and a greedy ensemble construction over the models produced by a standard Bayesian optimization.”