On 30 December 2024 the National Data Administration (NDA) released the first batch of “Commonly Used Terms in the Data Field”. The glossary fixes unified definitions for core concepts — data, data products and services, data processing, data security and data analysis. It looks modest next to a regulation, but vocabulary is the first layer of any market’s rulebook.
Why a glossary matters more than it looks
China’s data market has grown faster than its definitions. “Data product”, “data processing” and “data security” have been used across regulation, contracts and transactions with shifting meanings. When the same word means different things in a registration filing, an acquisition agreement and a security assessment, disputes and compliance gaps follow.
The glossary is the NDA’s attempt to anchor that vocabulary. It sits upstream of the market infrastructure the NDA has been building: the registration of data property rights and the rules for data circulation service institutions both presuppose that “data” and “data product” mean the same thing to everyone involved.
How it connects to the rest of the data framework
Definitions rarely stand alone. Once “data processing” is standardised, the obligations that attach to processing become easier to allocate — to the supplier, the platform or the buyer. The same is true of “data security”, a term that appears in sector rules and in security assessments alike.
The glossary also supports the supply side. Initiatives such as the national data bureau datasets guide and the industrial high-quality datasets action plan depend on a common understanding of what a dataset is and when it becomes a tradable product. For AI companies, whose training pipelines are themselves data-processing chains, that is not an abstract question.
Practical actions for AI companies
First, align internal definitions with the official ones. Contracts, data-processing agreements and internal policies that use “data product” loosely should be re-read against the glossary.
Second, use the glossary as a mapping tool. When a company registers data rights, files a dataset or negotiates a data transaction, matching its own terminology to the official definitions removes friction before it becomes a dispute.
Third, watch for the second batch. A glossary is a living document, and later tranches will cover terms such as data elements and data technology. The data technology innovation opinion signals where that vocabulary is heading.
In short, the glossary is not a compliance obligation in itself. It is the reference point against which later obligations will be measured.
This article is based on the National Data Administration’s Commonly Used Terms in the Data Field and public information and is for reference only; it does not constitute legal advice.