# Related-work ledger

Status: paper-local author ledger

Verified source date inherited from the binding dossier: 22 July 2026

This ledger records the results and system capabilities that Paper 4 may
attribute to prior work. It also records the novelty boundary. A citation
supports a claim only under the source's model and assumptions; it is never
evidence that CatDB implements the cited capability.

| Key | Established result or capability used by P4 | P4 boundary |
|---|---|---|
| `Spivak2012Functorial` | A schema functor induces \(\Sigma\), \(\Delta\), and \(\Pi\) data-migration functors with distinct variance. | Categories, functors, and these migration operators are not CatDB inventions. P4 must select an operator rather than infer data flow from an arrow. |
| `SpivakWisnesky2015Relational` | Functorial data migration has relational foundations and compositional behavior in the paper's admitted setting. | Composable categorical mappings and relational realization are established. |
| `SchultzWisnesky2017AlgebraicIntegration` | Algebraic data integration supplies formal integration constructions and constraints. | Algebraic integration is direct prior art, not a CatDB novelty. |
| `SchultzEtAl2017AlgebraicDatabases` | Algebraic databases develop categorical structure for schemas, instances, queries, and integration. | Categorical organization of databases is established. |
| `WisneskyEtAl2015Practice` | FQL demonstrates executable functorial migration in integration case studies. | P4 may not claim the first executable categorical integration workflow. |
| `BrownEtAl2019Computational` | CQL has been applied to constrained computational-science data integration. | A future CatDB evaluation must differ from merely running categorical integration. |
| `CQL2026` | CQL documentation exposes typed schemas, instances, mappings, transforms, constraints, and executable migration workflows. | Syntax and execution for categorical mappings are mature baselines. |
| `BernsteinMelnik2007Model` | Model management treats models and richer mappings as first-class values with operators such as match and compose. | “Mappings are first class” and composition alone are not contributions. |
| `SchultzEtAl2017ModelManagement` | Algebraic model management organizes established model-management operators categorically. | Categorical terminology around mapping operators is not enough for novelty. |
| `RahmBernstein2001Matching` | Schema matching methods generate candidate correspondences using multiple signals. | Candidate discovery is distinct from a validated, authorized executable mapping. |
| `MadhavanEtAl2001Cupid` | Cupid is a concrete generic schema-matching system. | P4 must compare matching assistance without claiming that typed mappings solve matching. |
| `FaginEtAl2005DataExchange` | Data exchange formalizes solutions, universal solutions, the chase, and certain answers. | Target construction and generated integration results require comparison with data exchange. |
| `FaginEtAl2005Composition` | Mapping composition has language-dependent expressiveness and closure boundaries. | “Mappings compose” must name the mapping language, equality, and supported fragment. |
| `ShethLarson1990Federated` | Federated database systems integrate heterogeneous, autonomous systems under explicit architectural choices. | Heterogeneous federation and autonomy are longstanding concerns. |
| `Wiederhold1992Mediators` | Mediators separate integration knowledge from underlying information sources. | Moving reusable meaning into a mediation layer is established prior art. |
| `KirkEtAl1995InformationManifold` | The Information Manifold uses a mediated information space for heterogeneous sources. | A shared integration domain and mapping topology are not new by themselves. |
| `Lenzerini2002DataIntegration` | Global-schema integration, GAV/LAV mappings, and query-answering obligations have a formal comparative foundation. | P4's mediated design must be evaluated against GAV, LAV, and related approaches. |
| `FaginEtAl2009Clio` | Clio creates schema mappings and generates data-exchange transformations. | Nonprocedural mapping creation and generated transformations are direct prior art. |
| `HaasEtAl2005ClioGrows` | Clio compiles abstract mappings into multiple executable output languages. | Semantic-to-physical compilation is not new in the broad sense. |
| `VassiliadisEtAl2001Arktos` | ARKTOS models, controls, and executes ETL processes, including operational concerns. | P4 cannot erase scheduling, cleaning, loading, quality, contingency, or execution work from its comparison. |
| `W3CR2RML` | R2RML standardizes relational-to-RDF mappings. | Declarative relational-to-semantic mappings are established. |
| `DimouEtAl2014RML` | RML generalizes declarative RDF mappings to heterogeneous source formats. | Heterogeneous-source mapping languages are direct research prior art. |
| `CalvaneseEtAl2017Ontop` | Ontop supports virtual ontology-based access to relational databases through rewriting. | Virtual semantic access and query rewriting require direct comparison. |
| `Dehghani2019DataMesh` | Practitioner data-mesh work frames domain ownership and decentralized data architecture. | This is an organizational/practitioner baseline, not independent CatDB systems evidence. |
| `Dehghani2020DataMeshPrinciples` | Data-mesh principles include federated governance and data as a product. | CatDB must state how its technical artifacts differ from organizational prescriptions. |
| `LookML2026` | LookML centralizes semantic definitions for analytical use. | Central semantic definitions and generated queries are mature product capabilities. |
| `DbtSemanticLayer2026` | The dbt Semantic Layer centralizes metric semantics across consuming tools. | Semantic-layer reuse is a product baseline, not a CatDB novelty. |
| `GoogleCloudELT2026` | Official documentation distinguishes ELT by loading before target-side transformation and records operational tradeoffs. | ELT remains a physical strategy; documentation is not comparative performance evidence. |

## Mandatory comparison dimensions

The manuscript and later review must compare the strongest relevant systems
along these dimensions:

1. source, domain, and consumer mapping orientation;
2. GAV, LAV, mediator, data-exchange, and functorial semantics;
3. admitted mapping language and closure under composition;
4. selected data-flow operator or query-rewrite semantics;
5. loss, partiality, ambiguity, synthesis, and identity authority;
6. virtual, federated, ELT, ETL, and materialized realization;
7. pair-specific exceptions and residual handwritten logic;
8. provenance and versioning of mappings and decisions;
9. topology, governance, and operational cost;
10. experimentally measured correct-repair surface.

## Citation discipline

All capabilities from this ledger are labeled `SUPPORTED BY PRIOR LITERATURE`.
Product and practitioner documentation supports only the documented
capability. No source establishes CatDB's proposed maintenance advantage.