基于BGE文本嵌入模型识别的数智赋能绿色技术创新体系研究
Research on the Digital-Intelligence-Empowered Green Technology Innovation System Identified via the BGE Text Embedding Model
DOI: 10.16538/j.cnki.fem.20260723.102
[1]谢家平,叶逸晨,谢吉青,周昱希.基于BGE文本嵌入模型识别的数智赋能绿色技术创新体系研究[J/OL].外国经济与管理,2026,48(09):25-50[2026-11-11].https://doi.org/10.16538/j.cnki.fem.20260723.102.
摘要
政府政策文件作为国家治理经验、制度认知、专家决策共识与实践反馈经验的制度化沉淀,是观察我国绿色技术创新体系政策设计逻辑和治理机制安排的重要文本入口,也是从中国实践中提炼绿色技术创新自主知识体系的重要素材。为了突破现有研究对绿色技术创新体系影响因素静态识别的局限,本文创新性地融合BGE文本嵌入模型和“输入—过程—输出”系统要素逻辑,构建“主题识别、语义向量化、方向性机制关系抽取、专家座谈校验与案例回访、政策语义有向网络生成”的分析框架。研究发现,当前政策设计呈现“重执行、轻反馈”的结构特征,目标导向、动力驱动、组织协同与过程运行之间已形成较为清晰的前向主干链条,而实践评估向目标优化、制度调整和过程改进的反馈连接相对薄弱。进一步看,数智赋能绿色技术创新超越了数智技术应用与政策工具供给的简单叠加,体现为多系统耦合形成的复杂治理体系。本文的贡献在于,从政策语义网络视角刻画数智赋能绿色技术创新体系的结构特征,提出识别复杂政策体系主干链条与反馈缺口的可操作方法,并通过人工智能语义识别与专家智慧校验的双源互动,结合案例实践回访的交叉印证,从中国政策实践中提炼具有本土解释力的绿色技术创新治理机制框架。
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.
关键词: 数智技术,绿色技术创新体系,中国自主知识体系