Abstract:

[Significance] Data in the agricultural industry chain is generated across production, processing, storage, logistics, sales, and regulation. It is characterized by dispersed stakeholders, long-chain processes, heterogeneous sources, and high sensitivity to privacy and business information. However, existing agricultural big data platforms, as well as single technologies such as blockchain-based traceability, privacy-preserving computation, and federated learning, are still insufficient to systematically address the bottlenecks of data ownership confirmation, controlled use, audit trail, and value collaboration in cross-entity data circulation. Therefore, the aim of this study is to address the demand for trusted cross-entity data circulation in the agricultural industry chain by proposing a trusted data space framework adapted to agricultural production, circulation, regulation, and service collaboration scenarios. It further clarified the layered architecture, key operating mechanisms, and implementation pathways of the framework, thereby providing a systematic reference for agricultural data sovereignty protection, controlled sharing, audit tracing, and value collaboration. [Progress] Based on the national standard Technical Architecture of Trusted Data Space, and considering the scenario characteristics of the agricultural industry chain, including multiple stakeholders, long-chain processes, limited computing capacity at edge nodes, data heterogeneity, and quality and safety supervision requirements, this study proposed an overall trusted data space framework for the agricultural domain. First, a four-layer architecture was designed, consisting of an infrastructure and data resource layer, a trusted data space core layer, a data capability support layer, and an agricultural industry chain application ecosystem layer. Then, four key operating logics underlying the framework were systematically explained. The first was a deep semantic interoperability mechanism based on ontology models, metadata, and knowledge graphs, which supported concept alignment, structural mapping, and contextual disambiguation of cross-entity data. The second was a controlled interaction mechanism based on standardized connectors, digital contracts, and usage control policies, which imposed fine-grained constraints on data usage purposes, invocation frequency, field scope, result output forms, and prohibited behaviors. The third was a value co-creation mechanism based on privacy-preserving computation and federated learning, which supported joint modeling, cross-domain risk analysis, and privacy-preserving production-sales matching without moving raw plaintext data out of local domains. The fourth was a dynamic trust mechanism based on blockchain-based evidence preservation and smart contracts, which enabled end-to-end auditable tracking of contract hashes, log hashes, result digests, and abnormal interception records. On this basis, a preliminary scenario-based description was provided using a self-developed integrated service platform for the trusted data space of the agricultural industry chain. The scenario description was based on approximately 120,000 food safety sampling and monitoring records from a city during 2023–2025. The scenario-based description showed that, in agricultural and food safety risk governance scenarios, the platform could support key trusted data space capabilities, including catalog management, semantic mapping, digital contract constraints, usage control enforcement, and on-chain evidence preservation and auditing. This scenario-based description provided a reference for the practical application of the framework. Finally, implementation pathways were proposed from four dimensions: institutional definition, technological research, ecosystem cultivation, and practical pilots. [Conclusions and Prospects] In response to the demand for trustworthy data circulation in the agricultural industry chain, this study proposed a domain-oriented trusted data space framework for agricultural scenarios. Supported by mechanisms such as connectors, semantic mapping, usage control, privacy-preserving computation, and on-chain auditing, the framework provided a reference for controlled data circulation and value collaboration among multiple stakeholders in the agricultural industry chain. In the future, with the further improvement of privacy-preserving computation components, data ownership definition, lightweight connectors, and data intermediary ecosystems, the trusted data space for the agricultural industry chain is expected to be increasingly integrated into the national data infrastructure system, providing sustained momentum for the efficient circulation of agricultural data elements, value release, and collaborative transformation of the whole industry chain.

Key words: agricultural industry chain data circulation, trusted data space, data element circulation, data connector, federated learning, agricultural informatization system