Recipe 21: Quiz answer adjudication #21

Recipe 21 of 60 · Level 2 (Easy) · Games & simulation · Choice

Use case

Judge whether a typed quiz response matches an accepted answer by meaning, with partial-match and review outcomes for ambiguous responses.

What to build

The folder recipes/21-quiz-answer-adjudication/ with notebook.ipynb, README.md, and fixtures/, meeting the recipe contract. Start from python tools/new_recipe.py 21 and the template in recipes/_template/.

Build notes

• Python checks normalised exact matches first. Only the rest go to Jev.
• Outcomes: match, partial_match, no_match, needs_review.
• Include misspellings, alternative names, over-specific answers, and answers that hedge between two options.

Decision design

• Choice. Options are a fixed set supplied by Python, including the fallback outcome this use case names. Show the choice, the per-option probabilities, and the confidence for individual examples before aggregating.

Level 2 scope

Context-sensitive matching, scoring, or verification with candidate sets and explicit uncertain or no-match outcomes.

Acceptance criteria

• recipes/21-quiz-answer-adjudication/notebook.ipynb runs top to bottom offline, with no key and no network, and its outputs are committed
• Live mode is opt-in (JEV_COOKBOOK_LIVE=1 with TYPESAFE_API_KEY) and uses the same question definitions as the offline run
• Fixtures are synthetic, include the hard cases named in the build notes, carry gold labels, and pass the fixture validator
• Every response fixture records its provenance, and the notebook states which mode it ran in
• Evaluation: Agreement with human adjudication, and how many answers never needed a model call.
• Candidate sets are assembled by Python and shown to the reader
• The uncertain or no-match outcome is exercised by several fixtures and reported separately in the evaluation
• No statement about Jev's quality, latency, or cost appears without recorded live inference; in a synthetic run, metrics are labelled as pipeline checks
• recipes/21-quiz-answer-adjudication/README.md says what the recipe teaches, how to run it, and what was and was not measured live
• The pull request changes only recipes/21-quiz-answer-adjudication/ and CI is green

Dependencies

• Blocked by Foundation: CI executes every notebook offline and validates fixtures #69 (CI executes every notebook offline and validates fixtures)
• Tracked in Epic: Level 2 · Easy recipes #76

Sources

Read these before writing the questions.

• S02 · TypeSafe AI: Primitives (Questions): Choice, Noul, Score, independent questions, and dependent follow-up calls.
• S03 · TypeSafe AI: Confidence: Distribution-derived confidence and task-specific thresholds informed by observed outcomes.
• S04 · TypeSafe AI: How to build with TypeSafe: Narrow semantic judgments integrated with workflows, deterministic logic, and side effects controlled by code.
• S07 · TypeSafe AI: Jev 1.13 jaggedness: Version-specific limitations affecting arithmetic, context selection, adversarial content, and generation.

Out of scope

• Changes to shared code, the catalog, or the root README. If the shared package is missing something, open a foundation issue and link it here.
• Real side effects of any kind. Actions are simulated.