MEASURING DIGITAL GREEN INNOVATION IN ENTERPRISES: A LIGHTWEIGHT TEXT MINING APPROACH BASED ON ANNUAL REPORTS

Abstract

Digital green innovation has become a critical pathway for Chinese enterprises to achieve sustainable development amid the convergence of the digital economy and green transition strategies. However, existing studies predominantly rely on structured indicators such as patents and financial data, which may fail to capture the strategic and contextual dimensions of firms’ innovation activities reflected in corporate disclosures. To address this gap, this study analyzes the annual reports of 1,285 Chinese listed manufacturing firms using web scraping and text mining techniques. Based on recent literature, a multidimensional indicator system containing 11 digital green innovation keywords is constructed to measure firms’ innovation activities. Empirical results reveal significant heterogeneity in keyword distribution across firms, with “artificial intelligence” and “Internet” appearing most frequently. Further analysis shows that heavily polluting enterprises exhibit significantly higher levels of green innovation, suggesting strong policy and regulatory incentives. By providing a scalable, text-based measurement framework, this study not only enriches the methodological toolkit for assessing digital green innovation, but also offers new empirical evidence on how firms disclose and prioritize innovation under environmental and digital pressures, thereby generating important implications for both policymakers and enterprise managers.