Explainable AI (XAI) for Constructing a Lexicon for Classifying Green Energy Jobs: A Comparative Analysis of Occupation, Industry, and Location Composition With Traditional Energy Jobs

Abstract

Growing concerns about climate change and environmental sustainability demand a shift from traditional to green energy. Insight into the different workforce profiles of green and traditional energy supply chains is critical to manage impacts to workers in traditional energy occupations and to identify what skills and occupations are needed in green energy jobs. The role and skills descriptions captured in big datasets of online job ads have the potential to deliver these insights but considerable time and effort is required to train algorithms to differentiate job ads pertaining to traditional and green energy occupations. In addition, algorithms usually need to be developed for each dataset of interest whereas lexicon-based approaches can be adapted relatively easily. In this study, we illustrate the use of explainable AI techniques (in combination with a set of initial keywords) to build a lexicon that can be used to identify (with 82% precision rate) traditional and green energy job ads across the supply chain (from energy production to energy policy, regulation and sales). Our study demonstrates the XAI-enhanced lexicon’s efficacy in uncovering differences in the industry, occupation, and geographic profile of traditional and green energy job ads. The findings provide valuable insights into the distributional consequences of the green energy transition by revealing where careful management of this transition is most needed whilst also illustrating the potential for XAI to support the development of precise and comprehensive lexicons.

Keywords

• Explainable AI (XAI)
• lexicon development
• green energy
• energy jobs classification
• workforce transition