On the Semantics of Generative SPARQL
Ratan Bahadur Thapa, Steffen Staab · 2026
Read On the Semantics of Generative SPARQL by Ratan Bahadur Thapa, Steffen Staab (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-23875.txt.
Introduction
Thapa and Staab give database queries the power to invent answers with a language model, without giving up the formal semantics that make queries analyzable, in a 2026 paper on database theory. SPARQL is the standard language for querying RDF knowledge graphs, and its semantics assume a fixed dataset. Applications increasingly want generated values produced on demand by a language model. Current practice bolts such calls on as external procedures, outside the algebra, so equivalence and termination questions go unanswered.
The authors add a generative operator, GenOp, whose evaluation calls a model and returns typed solution mappings, while the graph stays fixed. A compatibility relation generalizes equality to allow similarity based matching. Under bounded deterministic generation, tame fragments get fixpoint semantics, equivalence and rewrite rules are established, and query complexity matches standard SPARQL.
Catalog record
| CREATOR | Ratan Bahadur Thapa, Steffen Staab |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 183 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.23875