Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox


Scalable Link Prediction in Dynamic Networks via Non-Negative Matrix Factorization

Jul 23, 2016
Linhong Zhu, Dong Guo, Junming Yin, Greg Ver Steeg, Aram Galstyan



We propose a scalable temporal latent space model for link prediction in dynamic social networks, where the goal is to predict links over time based on a sequence of previous graph snapshots. The model assumes that each user lies in an unobserved latent space and interactions are more likely to form between similar users in the latent space representation. In addition, the model allows each user to gradually move its position in the latent space as the network structure evolves over time. We present a global optimization algorithm to effectively infer the temporal latent space, with a quadratic convergence rate. Two alternative optimization algorithms with local and incremental updates are also proposed, allowing the model to scale to larger networks without compromising prediction accuracy. Empirically, we demonstrate that our model, when evaluated on a number of real-world dynamic networks, significantly outperforms existing approaches for temporal link prediction in terms of both scalability and predictive power.

* Technical report for paper "Scalable Temporal Latent Space Inference for Link Prediction in Dynamic Social Networks" that appears in IEEE Transactions on Knowledge and Data Engineering 2016 


Share this with someone who'll enjoy it:

   Access Paper Source



Share this with someone who'll enjoy it: