高级检索

    基于图元学习的小样本虚假评论检测算法

    Graph-based Meta-learning Algorithm for Few-shot Fake Review Detection

    • 摘要: 基于图的虚假评论检测主要面临着如何在仅有少量正样本标注的属性图中,有效聚合图中不同关系的邻居信息,提高图表示学习对于异常节点的敏感性和泛化能力的挑战。针对此挑战,本文提出基于元学习的多信息融合图差异网络(Meta-MGDN)。通过构建多视图划分与多信息融合模块,充分挖掘用户、项目、评分的结构信息与属性信息,以实现网络对多方面信息的获取并挖掘评论节点之间的关系。本文设计多视图邻域差异聚合模块,合并邻域信息与自身-邻域差异信息,使网络同时关注节点之间的关联性与差异性,提高了网络对于异常节点的敏感性。最后,本文引入元学习框架,利用多个辅助网络增强目标网络学习小样本任务的经验,从而在小样本虚假评论检测场景下保持较高泛化能力。在真实公开评论数据集上进行实验表明,Meta-MGDN在基于图的虚假评论检测领域上效果优于先进的基线。

       

      Abstract: Graph-based fake review detection primarily faces the challenge of effectively aggregating neighbor information from different relations in an attributed graph with only a limited number of labeled positive samples. The goal is to enhance the sensitivity of graph representation learning to anomalous nodes and improve its generalization capability. To address this challenge, this paper proposes a meta-learning-based multi-information fusion graph difference network (Meta-MGDN). By constructing a multi-view partitioning and multi-information fusion module, the structural and attribute information of users, items, and ratings are fully explored to enable the network to capture multi-faceted information and uncover relationships between review nodes. This paper designs a multi-view neighborhood difference aggregation module that integrates neighborhood information with self-neighborhood difference information, enabling the network to simultaneously capture both the correlation and the differences between nodes, thereby enhancing its sensitivity to anomalous nodes. Finally, a meta-learning framework is introduced to simulate the few-shot problem using multiple auxiliary networks, which helps the network maintain a high generalization ability in the few-shot fake review detection scenario. Experiments on real-world datasets show that Meta-MGDN outperforms the state-of-the-art baselines in the field of graph-based fake review detection.

       

    /

    返回文章
    返回