A Generative Relational Domain-Specific Language for Evolving Social-Fact Records: Typed Assertions, Temporal Histories, and Governed Transitions

Public knowledge records need to preserve the distinction between a description, a person's commitment to it, an evidentiary assessment, and an authorized institutional action. This paper specifies GRK-DSL 0.1, a domain-extensible language for evolving social-fact records organized around Entity, Relation, State, Property, Process, and Event. Technical assertion and commit structures add attribution, explicit polarity, validity intervals, revision history, and controlled operational effects. The method combines requirements reconstruction, verified comparison with established formalisms, concrete and abstract syntax, static judgments, denotational and operational semantics, and counterexample-based assessment. Proposition identity separates intensional descriptions from complete-field extensional content. Four support states retain explicit opposition within an authorized historical view. A finite positive rule core derives virtual predicates through a least fixed point, while separate workflow commands produce admitted token receipts. Compatible history union preserves disagreement and exposes operational conflicts requiring explicit repair. The paper proves finite rule termination, grounded non-explosion, targeted withdrawal locality, and deterministic projection of compatible admitted receipts under stated assumptions. It identifies implementation-level type preservation and observational disclosure protection as further obligations. Complete grammar and worked encodings accompany bounded syntax, semantic-oracle, and structural-schema checks. These artifacts establish an inspectable language contract rather than a complete interpreter or deployment result. Comparative implementation and domain studies remain necessary to assess expressivity, usability, cost, and the language's contribution to responsible public knowledge evolution.