As an institutionalized condensation of national governance experience, institutional cognition, expert decision-making consensus, and practical feedback, government policy documents provide an important textual entry point for observing the policy design logic and governance mechanism arrangements of China’s green technology innovation system. They also constitute essential source materials for developing an independent knowledge system of green technology innovation grounded in Chinese practice. To move beyond the limitations of existing studies that mainly identify relevant influencing factors in a static manner, this paper integrates a text embedding model, namely BAAI General Embedding, with the Input-Process-Output system logic. It constructs an analytical framework that combines topic identification, semantic vectorization, directional mechanism relationship extraction, expert panel validation and case revisiting, and policy semantic directed network generation. The results show that the current policy system exhibits a structural pattern of “strong execution but weak feedback”. A relatively clear forward backbone chain has formed among goal orientation, driving forces, organizational coordination, and process operation, whereas feedback connections from practical evaluation to goal optimization, institutional adjustment, and process improvement remain comparatively weak. Furthermore, digital-intelligence-empowered green technology innovation extends beyond the simple combination of digital-intelligence technology applications and policy instrument provision. It represents a complex governance system formed through the coupling of multiple subsystems. The contribution of this paper lies in characterizing the structural features of the digital-intelligence-empowered green technology innovation system from the perspective of policy semantic networks, proposing an operational method for identifying the backbone chains and feedback gaps of complex policy systems, and extracting from China’s policy practice a governance mechanism framework for green technology innovation with local explanatory power.