CK-18 Numeric and attribution recomputation by code (numeric fidelity check; deterministic recomputation)
Extract every number, date, name and quotation from an output by code and confirm that each appears in, or can be derived from, the source, and produce a list of unmatched figures and names.
What it does
Extracts every number, date, name and quotation from an output with code and confirms that each appears in, or can be derived from, the source. Artefact: a list of unmatched figures and names for a person to resolve.
When to use it
Use when an output contains figures, dates, named people or attributed quotations that must be exactly right.
When it does not fit
Fits numerical claims and named attributions. Misfit: rounded, converted or aggregated figures, which need a defined tolerance.
Result format
The result lists every number, date, name and quotation extracted from the output, with a column showing whether it appears in or can be derived from the source. It ends with the list of unmatched figures and names for a person to resolve.
Requirements on the material
| Property | Requirement type | Requirement |
|---|---|---|
| Content type | Always applies | requires=text; requires=tabular |
| Unit of analysis | Always applies | requires=claim |
| Appraises source credibility | Applies in some cases | no |
Not restricted on: Structure, Variable types, Time structure, How the sample was chosen, Level of detail available, Number of sources, Whether sources can be joined, Amount of material, Known data problems, Producer interest, Source dependence, Origin, Document genre, Language, Personal information, Security classification, Licence.
| Minimum material | the source text or table and a stated tolerance for rounded or converted figures |
|---|---|
| Additional sources needed | none |
| Not for (rules) | none |
| Not for (text) | Wrong choice when figures are rounded, converted or aggregated and no tolerance is defined. A match shows the figure exists in the source, not that it is used correctly. |
Choosing and ordering
| Kind of technique | Verification technique, runs in step 6 |
|---|---|
| Phase tag | validate |
| Other phase tags | report |
| Kind of finding | contradiction |
| Question answered | how much; who |
| Strongest statement it supports | describe |
| Needs as input |
|
| Makes as output |
|
| Must exist first | None |
| Loop target | proceed with limit stated |
Library checks by stakes
Each stakes level includes the checks of the level below it. Checks that need a person apply at high stakes only, so the process leaves them out at lower stakes (decision 38). Repeatability: deterministic.
| Stakes | Check | Person involved | Counts as a different method |
|---|---|---|---|
| exploratory | input validation | none | no |
| decision-informing | input validation | none | no |
| decision-informing | code execution for arithmetic | none | no |
| decision-informing | spot check | optional | no |
| high-stakes | input validation | none | no |
| high-stakes | code execution for arithmetic | none | no |
| high-stakes | spot check | optional | no |
| high-stakes | human review sample | needed | no |
| high-stakes | independent review | needed | no |
Failure modes and limits
| Model suitability | model plus code |
|---|---|
| Typical model failure | Computes totals or percentages incorrectly, or attributes a quote to the wrong speaker. |
| Error sources | input data; data transformation |
| Certainty limits | Certainty drops when tolerances are loose or undefined (imprecision) and when matched figures are used in a different sense from the source (indirectness). |
Related techniques
- should follow PE-10 Schema-based structured extraction with an LLM (LLM (large language model) information extraction; form filling from documents): extracted amounts and dates are checked against the source
- should follow PE-12 Model summarisation of documents (abstractive summarisation; multi-document summarisation): figures in summaries are checked
- pairs with CK-17 Source-span and quotation check by code (quote-and-verify; extractive grounding; citation verification): quotation existence and figure matching cover different errors
Evidence and tag basis
Unconfirmed. No source names this technique. It applies the verification requirement in NIST AI 600-1, and the design is the note author's inference. (NIST AI 600-1)
Tag basis: input profile inference; selection inference; sequencing inference; robustness inference. All tag values are inference unless a source is named.