Large language models for tech mining and technology opportunity analysis: A diffusion of innovation perspective on detecting evolutionary stages of innovative practices using ChatGPT

Technology opportunity analysis (TOA) traditionally examines technological developments to identify opportunities, but typically conceptualizes these opportunities at the level of technologies or fields. This study advances practice-level TOA by conceptualizing technology-enabled innovative practices as technology-process-industry configurations and using their evolution to identify technology opportunities. It extends TOA by integrating diffusion of innovation theory and operationalizing its constructs to explain how opportunity spaces evolve. A seven-stage evolutionary model tracks how configurations emerge, diffuse and consolidate over time, identifying an opportunity window where configurations remain vacant, exploratory, emerging or promising. To investigate this perspective and respond to recent calls for generative AI in informetrics, the study develops an AI-enabled Tech Mining framework based on large-scale analysis of full-text scientific articles. ChatGPT-4o operates within an LLM+ architecture, combining prompt engineering with a rule-based expert system. The framework is benchmarked against expert coders, near-graduate students, a fine-tuned BERT classifier and alternative LLMs, and applied to 129,566 articles (2019–2024). The resulting dataset of 43,632 practices across 11 enabling technologies is analyzed using supervised machine learning to identify plausible vacant configurations. The study shows how controlled LLM workflows can make theory-informed opportunity analysis scalable and support a more managerial understanding of technology opportunities.