Link Prediction via Matrix Factorization

By Aditya Krishna Menon, Charles Peter Elkan. ECML/PKDD, 2011

575 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Statistical and Nonlinear Physics

View Paper: https://www.semanticscholar.org/paper/00d23e5c06f90bed0c9d4aec22babb2f7488817f

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 Statistical and Nonlinear Physics

Citations in the corpus

TensorCast: forecasting and mining with coupled tensors (2018, background): “The Netﬂix challenge sprung the creation of several latent factor models with differing structure and/or regularization terms for this task [19,25], but there were also several approaches which showed that using the age of the link could lead to improved predictions [20].”
Dropout Training of Matrix Factorization and Autoencoder for Link Prediction in Sparse Graphs (2015, methodology): “Among the large number of models proposed over the decade, Matrix Factorization (MF) is one of the most popular ones in network modeling and relational learning in general [27, 24, 14, 22].”
Beyond Instructional Videos: Probing for More Diverse Visual-Textual Grounding on YouTube (2020, methodology): “We now define a segment-level AUC metric, analogous to the intra-video AUC metric previously defined: it quantifies how readily individual ASR captions are temporally localized by the model within the same video (see Menon and Elkan (2011) for a description of different AUC variants).”
Opinion Fraud Detection in Online Reviews by Network Effects (2013, background): “Related to ratings on products, work on recommender systems (Koren 2009; Menon and Elkan 2011), aim for best prediction of future user ratings, but do not address the fraud problem.”
A Deep Architecture for Matching Short Texts (2013, background): “Many machine learning problems can be interpreted as matching two objects, e.g., images and captions in automatic captioning [11] [14], users and products in recommender systems, queries and retrieved documents in information retrieval.”
Negative Link Prediction in Social Media (2014, methodology): “In [19], several unsupervised algorithms based on low-rank matrix factorization are proposed.”
Scalable Temporal Latent Space Inference for Link Prediction in Dynamic Social Networks (2014, background): “Various approaches, including Bayesian inference [4], [5], multidimensional scaling (MDS) [6], matrix factorization [7], [8], [9], [10], [11], [12], and mixed-membership model [13], have been proposed to infer the static latent space representation of the observed network.”
Semantic Matching in Search (2014, background): “These include collaborative filtering [101, 1, 2, 102, 45, 149, 9, 44], image annotation [83, 76, 159], paraphrasing and textual entailment [62, 54, 139, 203, 55, 85, 165, 35], question answering [37, 126, 21, 205, 178, 130], cross-language information retrieval [64, 141, 140], short text conversati”
Joint Link Prediction and Attribute Inference Using a Social-Attribute Network (2014, background): “2006; Singh and Gordon 2008], matrix factorization and alignment [Menon and Elkan 2011; Scripps et al. 2009] based approaches have been proposed to leverage attribute information for link prediction, but they suffer from scalability issues.”
LRBM: A Restricted Boltzmann Machine Based Approach for Representation Learning on Linked Data (2014, methodology): “For this challenge, many existing models seek to represent link structures using selected statistics on networks, then combine selected statistics with node attributes.”
Recommender Systems (2016, unknown): “Although one can use matrix factorization methods within the link prediction framework of the previous section by using the approach in [432] as the base algorithm, it is more fruitful to design and optimize matrix factorization methods directly for trust networks.”
Deep Generative Models for Relational Data with Side Information (2017, methodology): “Among other methods that can incorporate side information in link prediction models beyond stochastic blockmod-els, (Menon & Elkan, 2011) presented a number of non-probabilistic approaches based on latent space models that directly use the side information in the link prediction ob-jective function…”
Popularity versus similarity in growing networks (2011, methodology): “The framework may thus be used for link prediction, a notoriously diﬃcult and important problem in many disciplines [29], with applications ranging from predicting protein interactions or terrorist connections to designing recommender and collaborative ﬁltering systems [30].”
Multitask Matrix Completion for Learning Protein Interactions Across Diseases (2016, unknown): “, 2001), matrix factorizationbased approaches (Koren et al., 2009; Menon and Elkan, 2011), and Bayesian approaches using graphical models ( Jin et al.”
Learning to Discover Social Circles in Ego Networks (2012, background): “Finally, there are works that model network data similar to ours [6, 17], though the underlying models do not form communities.”