Scholarlink

This work uses Graph Neural Network (GNN) for classifying public procurement containing collusion. For this, the GraphSAGE model is used to generate the embeddings of the nodes of a bipartite graph formed by bidders and tenders carried out in the period from 2010 to 2023 in 184 municipalities in the state of Ceará, Brazil. The data were obtained from the Court of Auditors of the State of Ceará (TCE) and added with political, socioeconomic and demographic information obtained from the Regional Electoral Court of Ceará (TRE-CE), the Brazilian Institute of Geography and Statistics (IBGE) and the Federal Revenue Service. The accuracy obtained by the supervised machine learning model was around 90%, using labeled data based on a criterion that considers bids containing a pair of bidders suspected of collusion due to the high frequency of participation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Public Procurement Collusion Identification Based on GraphSAGE Algorithm

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Marcos Leno Ferreira Pompeu,
  • Raimir Holanda Filho

摘要

This work uses Graph Neural Network (GNN) for classifying public procurement containing collusion. For this, the GraphSAGE model is used to generate the embeddings of the nodes of a bipartite graph formed by bidders and tenders carried out in the period from 2010 to 2023 in 184 municipalities in the state of Ceará, Brazil. The data were obtained from the Court of Auditors of the State of Ceará (TCE) and added with political, socioeconomic and demographic information obtained from the Regional Electoral Court of Ceará (TRE-CE), the Brazilian Institute of Geography and Statistics (IBGE) and the Federal Revenue Service. The accuracy obtained by the supervised machine learning model was around 90%, using labeled data based on a criterion that considers bids containing a pair of bidders suspected of collusion due to the high frequency of participation.