At online retail platforms, it is crucial to actively detect risks of fraudulent transactions to improve our customer experience, minimize loss, and prevent unauthorized chargebacks. Traditional rule-based methods and simple feature-based models are either inefficient or brittle and uninterpretable. The graph structure that exists among the heterogeneous typed entities of the transaction logs is informative and difficult to fake. To utilize the heterogeneous graph relationships and enrich the explainability, we present xFraud, an explainable Fraud transaction prediction system. xFraud is composed of a predictor which learns expressive representations for malicious transaction detection from the heterogeneous transaction graph via a self-attentive heterogeneous graph neural network, and an explainer that generates meaningful and human understandable explanations from graphs to facilitate further process in business unit. In our experiments with xFraud on two real transaction networks with up to ten millions transactions, we are able to achieve an area under a curve (AUC) score that outperforms baseline models and graph embedding methods. In addition, we show how the explainer could benefit the understanding towards model predictions and enhance model trustworthiness for real-world fraud transaction cases.

中文翻译：

---

xFraud：异构图上可解释的欺诈交易检测

在在线零售平台上，至关重要的是积极检测欺诈性交易的风险，以改善我们的客户体验，最大程度地减少损失并防止未经授权的退款。传统的基于规则的方法和简单的基于特征的模型要么效率低下，要么脆弱且无法解释。事务日志的异构类型实体之间存在的图结构信息丰富，难以伪造。为了利用异构图形关系并丰富可解释性，我们介绍了xFraud，这是一种可解释的欺诈交易预测系统。xFraud由预测器组成，该预测器通过自注意异构图神经网络从异构交易图中学习恶意交易检测的表示形式，以及一个解释器，该解释器从图形生成有意义且易于理解的解释，以促进业务部门的进一步处理。在xFraud在两个具有多达一千万笔交易的真实交易网络上进行的实验中，我们能够获得曲线下的面积（AUC）得分，其性能优于基线模型和图形嵌入方法。此外，我们展示了解释器如何使对模型预测的理解受益，并增强了现实世界中欺诈交易案例的模型可信度。