A Network-Based Approach to Understand Potentials for On-the-Job Training in Construction Professions

Recent studies indicated that automation will disproportionately affect middle-skill jobs and intensify the polarization in the labor market. A major challenge for middle-skill workers is that they may need to transition to different jobs during their professional life. Lifelong learning is a potential solution to help people transition to new jobs when necessary. One of the approaches to promote lifelong learning is on-the-job training that helps workers learn new skills through interaction with other professions. However, not all jobs offer the same level of potential for this approach. Understanding connections among construction jobs based on their required skills, knowledge, and abilities is the first step toward designing a systematic plan for on-the-job training. To address this need, we use network science to analyze the Occupational Information Network (O*NET) data developed by the US Department of Labor. A multipartite network of 84 construction jobs, 35 skills, 52 abilities, and 33 knowledge areas is created to calculate and analyze three network indices including degree centrality, eigenvector centrality, and closeness centrality to understand potential capacity of learning through interaction with other professions for each job.