基于聚类分析的串标审计模型建设研究

Study on Building of the Colluded Bidding Audit Model Based on Cluster Analysis

审计工作可以利用机器学习技术,提升审计的质量和效率,逐步向智能化发展。使用聚类分析法建设识别串标组合的审计模型,提取工程招投标系统的承包商信息一中标结果历史数据,计算节点互信息,挖掘投标方行为相关性,在关联投标方较多的条件下,根据投标人的中标概率,将其分为不同的投标行为异常组,可用于最大范围地发掘串标舞弊行为的线索。

Auditing can use the machine learning technique to refine its quality and efficiency and gradually develop into intelligence.The cluster analysis method is used to build the audit model that identify the colluded bidding,extract the contractor information of the project bidding system-the historical data of bid results and the mutual information of calculation nodes,and to tap into the bidder behavior correlation.In terms of more associated bidders and their winning probability,it can be divided into the outlier groups of different bidding behaviors to dig out the clues of colluded bidding frauds to the maximum extent.