A Node Pair Entropy Based Similarity Method for Link Prediction in Transportation Networks
Link prediction is a challenging problem. It is an approach to determine the possibility of potential or missing link between node pairs in a network. Researches on transportation network’s link prediction are mainly about travel time prediction, path prediction, traffic flow prediction, congestion prediction and so on. However, current studies are restrained by direction of the link or a new route. To solve this problem, a node pair entropy based similarity method for link prediction is proposed. Firstly, the initial state of all nodes in the node pair are initialized. Then, the influence weights of upstream node to lower nodes and the feedback state are determined. So the uncertainty degree of a path is obtained. Finally, the link prediction of the unconnected node pair is measured by node pair entropy. This method differentiates the roles of different nodes, and the connection between the common points is considered. It becomes a good solution for transportation network’s link prediction.
KeywordsTransportation networks Link prediction Node pair entropy Similarity-based method
This paper is supported by The Chinese the State 13 Five-year Scientific and Technological Support Project (2016YFB1200402), The Big-Data Based Beijing Road Traffic Congestion Reduction Decision Support Project (PXM2016014212000036) and The Project of The Innovation and Collaboration Capital Center for World Urban Transport Improvement (PXM2016014212000030).
- 7.Li RH, Yu JX, Liu J (2011) Link prediction: the power of maximal entropy random walk. In: ACM International conference on information and knowledge management, ACM, pp 1147–1156Google Scholar
- 10.Simmons R, Browning B, Zhang Y et al (2006) Learning to predict driver route and destination intent. In: IEEE intelligent transportation systems conference, IEEE, pp 127–132Google Scholar
- 11.Mowshowitz A (1967) Entropy and the complexity of graphs. University of Michigan, pp 225–240Google Scholar
- 13.Xu Z, Pu C, Sharafat RR et al (2016) Entropy-based link prediction in weighted networks. Chinese Physic b English 1:588–594Google Scholar