SLIND: Identifying Stable Links in Online Social Networks
- 2.4k Downloads
Link stability detection has been an important and long-standing problem in the link prediction domain. However, it is often easily overlooked as being trivial and has not been adequately dealt with in link prediction . In this demo, we introduce an innovative link stability detection system, called SLIND (Stable LINk Detection), that adopts a Multi-Variate Vector Autoregression analysis (MVVA) approach using link dynamics to establish stability confidence scores of links within a clique of nodes in online social networks (OSN) to improve detection accuracy and the representation of stable links. SLIND is also able to determine stable links through the use of partial feature information and potentially scales well to much larger datasets with very little accuracy to performance trade-offs using random walk Monte-Carlo estimates.
KeywordsLink stability Graph theory Online social networks Hamiltonian Monte Carlo (HMC)
This research is partially supported by National Science Foundation of China (No. 61672039, No. 61772034, No. 61503092) and Guangxi Key Laboratory of Trusted Software (No. kx201615).
- 1.Ozcan, A., Oguducu, S.G.: Multivariate temporal link prediction in evolving social networks. In: International Conference on Information Systems (ICIS-2015), pp. 113–118 (2015)Google Scholar
- 2.Mengshoel, O.J., Desai, R., Chen, A., Tran, B.: Will we connect again? Machine learning for link prediction in mobile social networks. In: Eleventh Workshop on Mining and Learning with Graphs, Chicago, Illinois, pp. 1–6 (2013)Google Scholar
- 3.Sol-Dickstein, J., Mudigonda, M., DeWeese, M.R.: Hamiltonian Monte Carlo without detailed balance. In: Proceedings of the 31st International Conference on Machine Learning (JMLR), vol. 32 (2014)Google Scholar
- 4.Ryohei, R.: Semi-supervised graph embedding approach to dynamic link prediction. arXiv preprint arXiv:1610.04351 (2016)