Journal of Computer Applications ›› 2025, Vol. 45 ›› Issue (8): 2566-2571.DOI: 10.11772/j.issn.1001-9081.2024081117

• Artificial intelligence • Previous Articles    

Combining preprocessing methods and adversarial learning for fair link prediction

Yifeng PENG, Yan ZHU()   

  1. School of Computing and Artificial Intelligence,Southwest Jiaotong University,Chengdu Sichuan 611756,China
  • Received:2024-08-08 Revised:2024-11-04 Accepted:2024-11-07 Online:2024-11-19 Published:2025-08-10
  • Contact: Yan ZHU
  • About author:PENG Yifeng, born in 2000, M. S. candidate. His research interests include link prediction, fairness in algorithm.
  • Supported by:
    Sichuan Provincial Science and Technology Program(2019YFSY0032)

结合预处理方法和对抗学习的公平链接预测

彭一峰, 朱焱()   

  1. 西南交通大学 计算机与人工智能学院,成都 611756
  • 通讯作者: 朱焱
  • 作者简介:彭一峰(2000—),男,四川广安人,硕士研究生,主要研究方向:链接预测、算法公平性
  • 基金资助:
    四川省科技计划项目(2019YFSY0032)

Abstract:

Link prediction is a crucial task in network analysis that explores interactions between entities and forecasts new potential relationships in evolving networks. However, link prediction may generate biases, especially concerning links between entities with sensitive attributes. For example, the issue of “filter bubble”, which amplify the isolation of publicly accessible information and reduce diversity for online users. To address these challenges, ALFLP (Adding Link and Adversarial Learning for Fair Link Prediction) method was proposed after combining preprocessing stage methods and processing stage methods for addressing “filtering bubble” issue from the perspective of algorithmic fairness. In preprocessing stage, by adding links to disadvantaged link groups, the difference in link density between different groups was reduced. In processing stage, the output of the preprocessing stage was input into the adversarial learning-based method, and the generator and discriminator were played together to promote more inter-group links, thereby alleviating the “filter bubble” situation. Experimental results on real datasets pokec_n and pokec_z show that compared with baseline methods such as Jaccard, ALFLP method has the AUC index improved by about 12 and 10 percentage points respectively, and the modred index improved by about 0.14 and 0.10 respectively. It can be seen that ALFLP method can achieve a good balance between fairness and prediction accuracy.

Key words: link prediction, fairness, filter bubble, preprocessing, adversarial learning

摘要:

链接预测是网络分析的一项重要任务,它研究个体之间的相互作用,并推断在未来不断发展的网络中可能出现的新关系;然而,在链接预测过程中可能存在偏见,特别是当涉及包含某些敏感属性的个体之间的链接时,如引发“过滤气泡”的效应,这一效应表现为在线用户公开信息的隔离程度加剧、多样性降低。针对上述问题,从算法公平性的角度缓解“过滤气泡”问题,将预处理阶段的方法和处理阶段的方法相结合,进而提出ALFLP (Adding Link and Adversarial Learning for Fair Link Prediction)方法。在预处理阶段,通过向劣势链接群组添加链接,减小不同群组的链接密度差异;在处理阶段,将预处理阶段的输出输入至基于对抗学习的方法,并通过生成器与鉴别器相互博弈,促进更多组间链接,缓解“过滤气泡”问题。在真实数据集pokec_n和pokec_z上的实验结果表明,与基线方法(如Jaccard)相比,ALFLP方法的AUC指标分别提高了约12和10个百分点,而modred指标分别提高了约0.14和0.10。可见,ALFLP方法能在公平性和预测准确性之间实现较好的权衡。

关键词: 链接预测, 公平性, 过滤气泡, 预处理, 对抗学习

CLC Number: