Study on Influencing Factors of Construction Workers' Unsafe Behavior Based on Text Mining

Area of Science:

• Construction Safety Management
• Occupational Health and Safety
• Text Mining in Accident Analysis

Background:

• Unsafe construction worker behavior is a primary cause of safety accidents.
• Accident investigation reports offer valuable insights for accident prevention.
• Effective knowledge sharing from these reports is crucial for safety improvement.

Purpose of the Study:

• To analyze construction accident investigation reports using text mining.
• To identify unsafe worker behaviors and their influencing factors.
• To understand the relationships between these influencing factors for better safety management.

Main Methods:

• Text mining techniques applied to 500 construction accident investigation reports from Shenzhen, China.
• Latent Dirichlet Allocation (LDA) topic model used for factor identification.
• Social Network Analysis (SNA) employed to determine factor importance and relationships.

Main Results:

• Key influencing factors identified: weak safety awareness, operating regulations, supervision dereliction of duty, equipment resources, and inadequate construction party supervision.
• Significant correlations found between weak safety awareness and supervision dereliction.
• Other identified correlations include equipment resources with poor construction environment, organization/coordination with inadequate supervision, and operating regulations with hidden dangers investigation.

Conclusions:

• The study enhances the theoretical understanding of construction worker unsafe behavior.
• Identified key factors provide targeted control directions for construction safety managers.
• Findings support effective curbing of unsafe behaviors and improvement of overall safety management levels.