The digital era has fundamentally transformed the underlying logic of scientific research and innovation. Given China’s relative resource constraints on a per capita basis, it is particularly essential to establish an efficient, collaborative, and market-oriented green innovation system. Rooted in China’s regional context, this paper integrates the classical grounded theory with the machine learning method to analyze multi-phase government policy texts, aiming to uncover the intrinsic structure and operational mechanisms of the green innovation system.

This paper firstly performs manual coding and grounded analysis based on two overarching policy documents: the Guiding Opinions on Establishing a Market-Oriented Green Technology Innovation System and the Implementation Plan for Further Improving the Market-Oriented Green Technology Innovation System (2023-2025), jointly issued by the National Development and Reform Commission (NDRC) and the Ministry of Science and Technology (MOST). These documents provide foundational guidance for the construction of China’s green innovation system. Through this analysis, it identifies six core systems: goal orientation, dynamic drive, organizational collaboration, process operation, institutional guarantee, and practical evaluation. Secondly, this paper analyzes a large sample of local implementation rules derived from these two guiding policies. It innovatively incorporates LDA topic modeling and hierarchical clustering algorithms into the grounded analysis process. A large-scale data analysis of 1,382 local implementation documents is conducted, extracting 32 topics and applying hierarchical clustering to validate and refine the grounded coding. The results show strong consistency between machine-generated clusters and manual coding, confirming the rationality and structural coherence of the six systems.Thirdly, this paper proposes a novel “diamond model of a market-oriented green innovation system”. This model reinforces a development philosophy centered on enterprises, driven by the market and guided by the government. It employs a chain transmission mechanism of “Sci-Tech Hubs-Sci-Tech Platforms-Sci-Tech Networks” to propel the innovation process, thereby achieving progressive enhancements in technological autonomy, shared services, and the synergistic integration of technological and industrial innovation.Finally, this paper explores key strategies for digitally empowering the green innovation system. On the one hand, it utilizes big data technology to accurately identify supply-demand matching in the green technology market and facilitate collaborative innovation clusters. On the other hand, it constructs a sci-tech innovation network cluster system to leverage the aggregation and bridging functions of platform enterprises, enabling deeper integration of technological and industrial innovation. The study preliminarily reveals the parallel, overlapping, and dynamically adaptive characteristics of the green innovation system in the digital era, offering a systematic and structured analytical framework for the innovation theory. It also provides practical insights for policymakers in policy mixes and path design to establish a market-oriented green technology innovation system, thereby supporting the efficient translation of sci-tech achievements and enhancing industrial leadership.