.png){kind=icon}
.png){kind=icon}
Return
Network structure guided multi-objective optimization approach for key entity identification
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2023.111115.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Multi-objective key entity identification problems have attracted the attention of researchers due to their potential real-world applications, which single-objective approaches simply cannot provide. However, existing multi-objective optimization models seldom consider the heterogeneous propagation costs of key entities. In addition, most multi-objective algorithms are directly used to solve these models, leading to poor performance since they easily fall into local optimal solutions and cause isolated points on the Pareto frontier. This study aims to address the multi-objective key entity identification problem by maximizing the propagation scale while simultaneously minimizing the heterogeneous propagation costs to obtain a more realistic model. A network structure guided approach was designed to solve the formulated model. This consisted of two new strategies, i.e., the best candidate selection strategy and the layered crossover operator strategy, to enhance candidates during the evolution process and to obtain more complete Pareto solutions. Finally, experiments conducted using synthetic and real-world networks show that the proposed approach improves upon five commonly studied multi-objective algorithms and achieves superior performance compared to similar approaches on the Pareto frontier across a number of metrics.
Keywords:
Key entity
Multi-objective
Network structure
Optimization approach
Pareto frontier
Journal
IF:
6.6
Papers:
1.4W
Citations:
4.8W