Frontiers in Physics (Oct 2022)

Role of degree and weighted coreness based on endpoints in link prediction

  • Jiaqi Hao,
  • Zheng Li,
  • Zhanhe Wu,
  • Jinming Ma

DOI
https://doi.org/10.3389/fphy.2022.1016535
Journal volume & issue
Vol. 10

Abstract

Read online

Many researchers propose link prediction models based on node similarity. Among all models, researchers found that the endpoint influence plays an important role in evaluating the similarity between endpoints. For endpoint influence, we consider that an endpoint possessing a large and extensive maximum connected subgraph can strongly attract other nodes. After thorough research, we found that the coreness can describe the aggregation degree of neighbors and the endpoint degree may be used to describe the largest connected subgraph of an endpoint. In order to create a model, we repeat our experiments on eight real benchmark datasets after combining endpoint degree and weighted coreness. The experimental results illustrate the positive role of synthetical endpoint degree and weighted coreness for measuring endpoint influence in link prediction.

Keywords