Proceedings Article10.1145/2043932.2043972
Multi-value probabilistic matrix factorization for IP-TV recommendations
Yu Xin,Harald Steck +1 more
- 23 Oct 2011
- pp 221-228
25
TL;DR: This paper presents the first approach that takes into account also negative feedback when training on implicit feedback data, and sheds light on the implicit assumptions underlying the most successful approach to IP-TV (Internet Protocol Television) recommendations.
read more
Abstract: Matrix factorization (MF) has evolved as one of the most accurate approaches to collaborative filtering. In this paper, we extend the probabilistic MF framework as to account for multiple observations for each matrix element. This significantly improves the accuracy of recommender systems in several areas: (1) aggregation of ratings concerning items organized hierarchically, (2) (partial) compensation for the selection bias in the observed data by using an appropriate prior with virtual data points, and (3) improved recommendations of TV shows. While our framework applies to explicit and implicit feedback data, we outline in detail the latter application in this paper: we present the first approach that takes into account also negative feedback when training on implicit feedback data. Moreover, we shed light on the implicit assumptions underlying the most successful approach to IP-TV (Internet Protocol Television) recommendations in [Hu et al. 2008]. In our experiments, we obtain significant improvements over the existing approach.
read more
Chat with Paper
AI Agents for this Paper
Find similar papers on Google Scholar, PubMed and Arxiv
Write a critical review of this paper
Analyze citations of this paper to find unaddressed research gaps
Citations
Evaluation of recommendations: rating-prediction and ranking
Harald Steck
- 12 Oct 2013
TL;DR: This paper examines both rating prediction and ranking approaches in detail, and finds that the dominating difference lies instead in the training and test data considered: rating prediction is concerned with only the observed ratings, while ranking typically accounts for all items in the collection, whether the user has rated them or not.
284
RecTime: Real-Time recommender system for online broadcasting
Yoojin Park,Jinoh Oh,Hwanjo Yu +2 more
TL;DR: RecTime proposes a real-time recommender system for online broadcasting called RecTime which considers time factors and preferences simultaneously, and significantly outperforms previous methods in terms of the accuracy on the recommendation time as well as the items.
41
•Dissertation
Recommender system performance evaluation and prediction: information retrieval perspective
Alejandro Bellogín Kouki
- 01 Jan 2012
TL;DR: This thesis investigates the definition and formalisation of performance predic- tion methods for recommender systems, and studies adaptations of search performance predictors from the Information Retrieval field, and proposes new pre- dictors based on theories and models from Information Theory and Social Graph Theory.
22
User identification for enhancing IP-TV recommendation
Zhijin Wang,Liang He +1 more
TL;DR: This paper proposes an algorithm that first identifies users in accounts, then provides recommendations for each user, and results show that the proposed algorithm gives substantially better results than previous approaches.
19
No, That's Not My Feedback: TV Show Recommendation Using Watchable Interval
Kyung-Jae Cho,Yeon-Chang Lee,Kyungsik Han,Jaeho Choi,Sang-Wook Kim +4 more
- 08 Apr 2019
TL;DR: This work investigates the inherent characteristics of implicit feedback given in the TV show domain, and proposes a novel framework based on collaborative filtering that effectively solves the challenges and significantly outperforms other existing state-of-the-art methods.
16
References
•Proceedings Article
Probabilistic Matrix Factorization
Andriy Mnih,Ruslan Salakhutdinov +1 more
- 03 Dec 2007
TL;DR: The Probabilistic Matrix Factorization (PMF) model is presented, which scales linearly with the number of observations and performs well on the large, sparse, and very imbalanced Netflix dataset and is extended to include an adaptive prior on the model parameters.
Factorization meets the neighborhood: a multifaceted collaborative filtering model
Yehuda Koren
- 24 Aug 2008
TL;DR: The factor and neighborhood models can now be smoothly merged, thereby building a more accurate combined model and a new evaluation metric is suggested, which highlights the differences among methods, based on their performance at a top-K recommendation task.
Collaborative Filtering for Implicit Feedback Datasets
Yifan Hu,Yehuda Koren,Chris Volinsky +2 more
- 15 Dec 2008
TL;DR: This work identifies unique properties of implicit feedback datasets and proposes treating the data as indication of positive and negative preference associated with vastly varying confidence levels, which leads to a factor model which is especially tailored for implicit feedback recommenders.
Restricted Boltzmann machines for collaborative filtering
Ruslan Salakhutdinov,Andriy Mnih,Geoffrey E. Hinton +2 more
- 20 Jun 2007
TL;DR: This paper shows how a class of two-layer undirected graphical models, called Restricted Boltzmann Machines (RBM's), can be used to model tabular data, such as user's ratings of movies, and demonstrates that RBM's can be successfully applied to the Netflix data set.
Related Papers (5)
Yifan Hu,Yehuda Koren,Chris Volinsky +2 more
- 15 Dec 2008
Andriy Mnih,Ruslan Salakhutdinov +1 more
- 03 Dec 2007