Listening to the Workforce: Measuring Construction Worker Safety Attitudes from Social Media Discourse Using LLMs
arXiv:2606.04450v1 Announce Type: new Abstract: Worker safety attitudes are key determinants of whether protective practices are applied or bypassed on construction sites. Yet measuring them at scale has remained out of reach. Safety attitudes are multidimensional, vary across topics, and surface most candidly in workers' own conversations. This study created and validated the Construction Safety Attitude Framework (CSAF), which integrates two components: a theory-grounded structure that characterizes safety attitudes along eight dimensions, and an operational codebook for measuring them in wo
The proliferation of social media discourse and advancements in large language models make it possible to analyze human sentiment at scale, which was previously impractical for nuanced topics like safety attitudes.
Understanding worker safety attitudes at scale allows for proactive intervention and more effective safety protocols, potentially reducing accidents and improving occupational health in high-risk sectors.
The ability to quantify previously unmeasurable, multi-dimensional safety attitudes in construction introduces a new paradigm for risk assessment and safety management, moving beyond post-incident analysis.
- · Construction companies
- · Safety technology providers
- · Insurance companies
- · Workers
- · Companies with poor safety records
Improved safety outcomes and reduced incident rates on construction sites due to better insight into worker attitudes.
Development of new AI-powered tools and services for industrial safety monitoring and predictive analytics based on CSAF.
Potential for similar safety attitude measurement frameworks to be applied across other high-risk industries, leading to broader improvements in industrial safety standards and practices.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL