Semantics ladder

How much does an LLM gain from being told what data means?

The experiment

Goals
Method

Credit ratings

Research analytics

The results

All data is invented: the companies, LEIs, ratings, works, people and institutions. Nothing here is any rating agency's or bibliographic database's data or definitions.

Last updated 6 November 2026.

Q27, run 2026-11-04-1711-sonnet

← back to the run

How is Corvenia Power's 2032 note rated?

Tests: Instrument versus issuer: the subordinated note sits below the issuer.

Expected: rating = A-

✅ R1 csv run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R1 csv run 2 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R1 csv run 3 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R1 csv run 4 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R1 csv run 5 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R2 rdf run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R2 rdf run 2 graded by code

A- (A42, I06).

JSON: {"rating": "A-"}

✅ R2 rdf run 3 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R2 rdf run 4 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R2 rdf run 5 graded by code

A-, from A42 on I06.

JSON: {"rating": "A-"}

✅ R6 rdf+ontology run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R6 rdf+ontology run 2 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R6 rdf+ontology run 3 graded by code

A-, from A42.

JSON: {"rating": "A-"}

✅ R6 rdf+ontology run 4 graded by code

A-. I06 is rated A- in A42.

JSON: {"rating": "A-"}

✅ R6 rdf+ontology run 5 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R5 csv+ontology run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R5 csv+ontology run 2 graded by code

A-, from A42.

JSON: {"rating": "A-"}

✅ R5 csv+ontology run 3 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R5 csv+ontology run 4 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R5 csv+ontology run 5 graded by code

A-, from A42.

JSON: {"rating": "A-"}

✅ R4 csv+markdown run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R4 csv+markdown run 2 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R4 csv+markdown run 3 graded by code

A-, from A42.

JSON: {"rating": "A-"}

✅ R4 csv+markdown run 4 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R4 csv+markdown run 5 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R3 csv+padding run 1 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R3 csv+padding run 2 graded by code

A-, from A42.

JSON: {"rating": "A-"}

✅ R3 csv+padding run 3 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R3 csv+padding run 4 graded by code

A- (A42).

JSON: {"rating": "A-"}

✅ R3 csv+padding run 5 graded by code

A- (A42).

JSON: {"rating": "A-"}