基于集体评分行为的在线用户声誉识别

Identifying Online User Reputation in Terms of Collective Rating Behaviors

在线评级系统中,用户声誉是基于历史评级行为而形成的信任度,是区分正常用户与恶意用户的关键指标。未明确定义声誉可能导致系统易受恶意攻击。因此,清晰界定用户声誉对于有效排除恶意用户和准确评估目标质量极为重要。本文从复杂网络的视角全面回顾了用户声誉评估方法,特别着重于不同计算模型的探讨,包括迭代模型、基于群组的模型和基于特殊分布的模型。此外,本文对现有文献进行了系统分析,说明了各方法的优势和局限性,并对未来研究方向提出展望。

In online rating systems,user reputation is defined as a degree of trust formed based on historical rating behaviors and serves as a critical indicator to differentiate between genuine and malicious users.A lack of precise definition for reputation can render the system vulnerable to attacks by ma-licious entities.Thus,clearly defining user reputation is crucial for effectively excluding malicious users and accurately assessing the quality of the targets.This article comprehensively reviews methods for assessing user reputation from the perspective of complex networks,with a particu-lar focus on different computational models,including iterative models,group-based models,and models based on special distributions.Finally,this paper provides a systematic analysis of the ex-isting literature,illustrates the strengths and limitations of each method,and provides an outlook on future research directions.