Inconsistency in online consumer reviews (OCRs) may cause uncertainty and confusion to consumers when they make purchase decisions. However, there is a lack of a systematic and empirical investigation of review inconsistency in the literature. This research characterizes review inconsistency from multiple aspects, including rating-sentiment, content, and language, and proposes hypotheses about their effects on fake OCR detection by drawing upon deception and attitude-behavior consistency theories. We characterize review inconsistency with 22 features, and test the hypotheses with machine learning models developed for fake OCR detection. Our empirical evaluation results using real OCRs not only confirm the presence of review inconsistency, but also demonstrate significant positive effects of review inconsistency on the performance of fake OCR detection. The research findings have important implications for improving the effectiveness of consumer decision making and the trustworthiness of OCRs.

在线消费者评论（OCR）中的不一致可能会导致消费者在做出购买决定时产生不确定性和困惑。但是，在文献中缺乏对系统评价不一致问题的系统和实证研究。这项研究从多个方面（包括评分情感，内容和语言）来描述评论的不一致，并通过运用欺骗和态度-行为一致性理论，提出了有关其对假OCR检测的影响的假设。我们使用22种功能来描述评论不一致的特征，并使用针对伪造OCR检测而开发的机器学习模型来检验假设。我们使用实际OCR进行的经验评估结果不仅确认了评论不一致，而且还证明了审查不一致对伪造OCR检测性能的显着积极影响。研究结果对提高消费者决策的有效性和OCR的可信赖性具有重要意义。