Open AccessExploiting Node Similarity Based on Graphical Markov Models for Link Prediction
Author(s) -
Jing Sun,
Zhijie Lin
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1631/1/012149
Subject(s) - computer science , link (geometry) , similarity (geometry) , benchmark (surveying) , data mining , exploit , node (physics) , markov chain , graphical model , perspective (graphical) , margin (machine learning) , machine learning , artificial intelligence , engineering , image (mathematics) , computer network , computer security , geodesy , structural engineering , geography
Link prediction has been attached more attention in recent years. In this paper, we develop a link prediction method which has a unique perspective on using network structures. The key idea is to exploit the relationship information based on Graphical Markov models (GMM) for designing similarity indices. Specifically, networks with GMM were modeled to capture the relational influence of nodes by taking multi-hop neighbors into consideration. Then, link ties are measured for supporting relationship prediction based on the theory of weak and strong ties. Moreover, the proposed method can be used to predict the emergence of future relationships between the nodes. Finally, empirical studies on real-world dataset demonstrate that the benchmark of the proposed method improves with a significant margin compared with other methods.