基于行为模式的用户声誉度量方法研究

在线评级系统由于水军和恶意打分者的存在而无法对物品给出客观评价，因此，建立一个基于打分行为的声誉度量模型对于在线评级系统的健康发展至关重要。现有的用户声誉度量方法仅依靠用户评分和商品质量之间的差异进行计算，忽略了用户的行为模式。将用户的评分偏差和行为模式相结合，提出了一种新的声誉度量方法，该方法不仅考虑了用户打分频率的极值，还考虑了用户打分总次数。在两个实证数据集上的实验结果表明，新方法对随机打分的识别准确率相较于经典算法最高可以提高17%，对于解决冷启动和鲁棒性问题具有更好的表现。

Measurement of user reputation via users' behavior patterns

Online rating systems fail to measure qualities of items due to attacks given by unfair raters, and it is crucial for the health of online rating systems to establish a reputation ranking system to identify unfair raters. The existing user reputation measurement methods only take into account the difference between the user''s rating information and the item qualities, regardless of users'' rating behavior patterns. Combining users'' rating bias and behavioral patterns, a new reputation ranking method BPR was proposed, and the model considered not only the extremes of user rating frequency, but also the total number of user ratings. The extensive experimental results for empirical datasets show that, comparing with the classical method, the accuracy of the BPR method for identifying random ratings by large-scale users could be improved by up to 17%, with better performance for cold start and robustness problems.