Governed Analytical Service
Analytical systems are good at producing numbers. A harder question is when a number is entitled to become an answer. This paper develops Governed Analytical Service (GAS) as an implementation-independent framework for that question. It begins from two gaps introduced in earlier work on Analytical Governance. The intent gap lies between a user's purpose and the governed analytical request that may stand for it. The servability gap lies between that request and an answer established under analytical law and current evidence. The first gap is closed when materially different governed interpretations have been resolved without the service taking a choice it does not own. The second is closed when a candidate realization has been shown to establish the requested analytical object. In both cases, capability is separated from authority: a plausible interpretation is not yet a request, and a computable result is not yet an answer. GAS therefore requires a governed request boundary, substantive analytical law, establishment evidence, independent adjudication, typed outcomes, and a serving boundary whose standing is respected in deployment. That boundary must also survive the serving protocol: a refusal should not carry a substitute value that a downstream agent can turn into the answer. GAS does not prescribe one query language, semantic layer, execution engine, or theory. The paper compares several common analytical architectures through these responsibilities and presents Columna only as one emerging implementation.
Authors
- Huayin Wang
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-10-04
- DOI
- https://doi.org/10.5281/zenodo.23150163
- Primary Topic
- Big Data and Business Intelligence
- Type
- preprint