Research on prediction of corruption behavior in Chinese construction bidding and tendering using UP technology

Purpose Since the 14th Five-Year Plan, the construction industry has entered a new phase of development, but corruption incidents in construction bidding and tendering have also occurred frequently. While corruption severely affects the sustainable development of the construction industry, investigating and governing corruption behavior holds great significance. This study aims to establish the user profile of corruption in construction bidding and tendering, predict corruption behavior and thereby take targeted anti-corruption measures. Design/methodology/approach This study constructs the label system based on corruption case data and employs user profile (UP) technology integrated with statistical analysis and the latent Dirichlet allocation (LDA) model to develop the user profile of corruption in construction bidding and tendering. Using the characteristics of the user profile, similar corruption subjects are identified and corruption behaviors in target subjects are predicted through the collaborative filtering recommendation (CF) algorithm based on behavior patterns of similar subjects, thereby enabling the formulation of targeted anti-corruption governance strategies. Findings Results indicate that the number of corruption cases fluctuates annually. Corruption is unevenly distributed regionally, with higher concentrations in third-tier and fourth-tier cities. Male subjects constitute the majority of corruption subjects, particularly among top leaders and key positions. Furthermore, subjects with aggressive risk preference constitute the majority. The primary corruption patterns involve taking bribes and collusion, with certain corruption patterns correlated with the position of subjects. The CF model achieves accurate prediction of corruption behavior in target subjects. Originality/value This study introduces UP technology into the field of corruption in construction bidding and tendering, conducting characteristic label analysis on corruption subjects. By exploring corruption profiles and behavior patterns from a micro-level perspective of corruption subjects, it contributes novel insights to this research field. From the perspective of practical significance, the research accurately identifies potential corruption subjects and behaviors and proposes effective corruption control strategies.