CLiMF: learning to maximize reciprocal rank with collaborative less-is-more filtering

By Yue Shi, Alexandros Karatzoglou, Linas Baltrunas, Martha A. Larson, Nuria Oliver, Alan Hanjalic. 2012

339 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, 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 Artificial Intelligence

In Information Systems

In Recommender Systems

Citations in the corpus

Recurrent Neural Networks with Top-k Gains for Session-based Recommendations (2017, neutral): “sentially a listwise loss function was introduced along with an alternating bundle method for optimization of the factors. Further ranking loss function for collaborative filtering were introduced in [18] [15] and [12]. Note that the fact that these loss functions work well in matrix factorization d”
Session-based Recommendations with Recurrent Neural Networks (2015, neutral): “Although the task can also be interpreted as a classiﬁcation task, learning-to-rank approaches (Rendle et al., 2009; Shi et al., 2012; Steck, 2015) generally outperform other approaches.”
Gaussian Ranking by Matrix Factorization (2015, background): “This is a diﬃcult optimization problem, as recommender systems typically predict a continuous-valued score for each item, based on which the items are then ranked for a given user.”
Joint Topic-Semantic-aware Social Recommendation for Online Voting (2017, methodology): “To mitigate this issue, many approaches have been proposed to utilize social network information in recommender systems [3, 6, 10, 12, 25, 29, 31, 32].”
StarSpace: Embed All The Things! (2017, methodology): “In the domain of recommendation, embedding models have had a large degree of success, starting from SVD (Goldberg et al. 2001) and its improvements such as SVD++ (Koren and Bell 2015), as well as a host of other techniques, e.g. (Rendle 2010; Lawrence and Urtasun 2009; Shi et al. 2012).”
Collaborative Denoising Auto-Encoders for Top-N Recommender Systems (2016, background): “, WARP [29, 30], CLiMF [21]).”
Social Collaborative Viewpoint Regression with Explainable Recommendations (2017, methodology): “CliMF Maximize reciprocal rank method for item ranking [37] LRMF List-wise learning to rank method for item ranking [36] NMF Non-negative matrix factorization [20] PMF Probabilistic matrix factorization [28] SoMF Trust propagation matrix factorization [15] TrMF Trust social matrix factorization [50]”
Optimizing top-n collaborative filtering via dynamic negative item sampling (2013, background): “This is essentially diﬀerent from most ranking-oriented CF algorithms [6, 7, 9] where ﬁrstly the target ranking is somehow smoothed strategies.”
Learning to rank for recommender systems (2013, methodology): “Typical examples of those methods are CoﬁRank [12], CLiMF [10] and TFMAP [9].”
Learning to rank recommendations with the k-order statistic loss (2013, neutral): “Methods like CofiRank [7], CLiMF [5], or Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citati”
Transfer Learning for Heterogeneous One-Class Collaborative Filtering (2016, background): “Most algorithms exploiting one-class feedbacks focus on homogeneous positive feedbacks with poitwise [1, 5], pairwise [10] and listwise [11] preference assumptions.”
A Generic Coordinate Descent Framework for Learning from Implicit Feedback (2016, background): “Both approaches are considered state-of-the-art and while CD outperforms BPR on certain datasets [8, 25, 15, 22, 26], BPR has been shown to work better on others [4, 17, 16, 8].”
Adaptive Bayesian personalized ranking for heterogeneous implicit feedbacks (2015, methodology): “Mathematically, they are defined as follows [19], Pre@k ¼ 1 U tej j P u2U te 1 k Pk p¼1d IuðpÞ 2 I teu , where U te is the set of users in the test data, I teu is the set of preferred items by user u in the test data, and dðxÞ is an indicator function with dðxÞ ¼ 1 if x is true and dðxÞ ¼ 0 otherwis”
Collaborative Filtering-Based Recommendation of Online Social Voting (2017, background): “Many OSNs now offer the social voting function, through which a user can share with friends her opinions, e.g., like or dislike, on various subjects, ranging from user statuses, profile pictures, to games played, Manuscript received October 22, 2014; revised August 19, 2016; accepted January 24,…”
Improving pairwise learning for item recommendation from implicit feedback (2014, methodology): “Recently, CLiMF [19] has been proposed for optimizing a matrix factorization model using implicit feedback datasets for the reciprocal rank measure.”
Scalable Recommendation with Hierarchical Poisson Factorization (2015, background): “• CliMF : Collaborative Less-is-More ﬁltering [34] maximizes mean reciprocal rank to improve the top-n predictive performance on binary relevance data sets.”
Transfer Learning for Behavior Ranking (2017, methodology): “For a listwise method, some loss function deﬁned on more than two (user, item) pairs is used to be optimized, which is usually designed to approximate some ranking-oriented evaluation metric [Shi et al. 2012].”