Modeling relationships at multiple scales to improve accuracy of large recommender systems

By Robert M. Bell, Yehuda Koren, C. Volinsky. Knowledge Discovery and Data Mining, 2007

411 citations recorded by the source corpus.

Subfields: Information Systems, Management Science and Operations Research, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/cae130eb00f79a415b4ba2f01dbb7019c4a4666d

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 Science and Operations Research

In Statistical and Nonlinear Physics

Citations in the corpus

Collaborative filtering with temporal dynamics (2010, methodology): “One of the more successful approaches to CF is based on a matrix factorization model [ 2 , 5, 10, 13, 18].”
Advances and Challenges in Conversational Recommender Systems: A Survey (2021, methodology): “Even interpreting user responses to such questions is considered a challenging task [6].”
A Distributed, Architecture-Centric Approach to Computing Accurate Recommendations from Very Large and Sparse Datasets (2011, background): “However, it does not explicitly adjust to users’ rating scales [11].”
Collaborative Filtering for Implicit Feedback Datasets (2008, background): “On the other hand, if u never consumed i, we believe no preference (pui = 0).”
Fused Matrix Factorization with Geographical and Social Influence in Location-Based Social Networks (2012, methodology): “To solve large-scale recommendation problems, matrix factorization is a promising tool due to its success in Netflix competition (Bell, Koren, and Volinsky 2007; Koren 2009).”
Regression-based latent factor models (2009, methodology): “This is evident in the plethora of methods that have been proposed in the context of the movie recommender systems.”
Finding your friends and following them to where you are (2012, background): “For instance, behavioral patterns of people taking taxis, rating movies, choosing cell phone providers, or sharing music are often best predicted by the habits of related people, rather than by the attributes of the individual such as age, ethnicity, or education [3, 24].”
Recommender Systems (2016, positive): “However, some of these dimensionality reduction techniques [71, 72, 309, 525] are specifically designed to improve the effectiveness and efficiency of neighborhood-based techniques.”
Collaborative filtering with temporal dynamics (2009, methodology): “1 The anatomy of a factor model One of the more successful approaches to CF is based on a matrix factorization model [2, 5, 10, 13, 18].”
A Comprehensive Survey of Neighborhood-based Recommendation Methods (2011, methodology): “This can help the user better understand the recommendation and its relevance, and could serve as basis for an interactive system where users can select the neighbors for which a greater importance should be given in the recommendation [4].”
Recommender systems with social regularization (2011, background): “Related recommendation techniques have been widely studied in research communities of information retrieval [5, 8, 12, 21, 36, 38, 39, 40], machine learning [28, 30, 31, 33, 41] and data mining [1, 3, 11, 14, 15].”
The BellKor solution to the Netflix Prize (2007, methodology): “All factorization models significantly benefit from recomputing user-factors in a neighborhood-aware fashion, in the spirit described in our KDD'2007 paper ([3], Sec.”
Factorization meets the neighborhood: a multifaceted collaborative filtering model (2008, background): “Hence, more recent works [4, 6, 9, 17, 18, 22] suggested modeling directly only the observed ratings, while avoiding overfitting through an adequate regularized model, such as:”
Personalized Recommendation Combining User Interest and Social Circle (2014, neutral): “It can be viewed as the first generation of RSes [7] with traditional collaborative filtering algorithms [8]– [12], [22]–[32] to predict user interest.”
Probabilistic factor models for web site recommendation (2011, background): “Recently, due to its efficiency in dealing with large datasets, several low-dimensional matrix approximation methods [1, 13, 21, 22, 23, 26] have been proposed.”
Scalable Coordinate Descent Approaches to Parallel Matrix Factorization for Recommender Systems (2012, methodology): “A similar procedure with the feature-wise update sequence is also used in [18] to avoid the over-fitting issue in recommender systems.”
fLDA: matrix factorization through latent dirichlet allocation (2010, methodology): “Of these, factorization models have become popular and have provided good performance in several real-world applications [25, 5].”