Understanding Jev: choice, score or probability?
Define the decision clearly before writing your first integration.
What each question type answers
The official Jev interface organizes questions around Choice, Score and Noul. Choice selects from supplied options, Score evaluates defined criteria, and Noul returns a truth-judgment value between 0 and 1.See official definitions
| Question type | What to ask | How business code uses it |
|---|---|---|
Choice | Which candidate applies? | Select a branch or route |
Score | Where does it fall on the defined scale? | Rank, compare or segment |
Noul | Does this statement hold? | Apply a threshold |
Split a big judgment into independent questions
Identify independent factors rather than bundling many requirements into one vague score. For example, classify a support ticket's department, urgency and need for a human separately, then let business code choose the next step. This is a design example, not a test result.
The official introduction also emphasizes atomic questions and combining answers in code.Read the official introduction
Correct types do not guarantee correct decisions
Check type conformance separately from factual correctness. Prepare human review or fixed rules for uncertain answers, and retain inputs, answers and final outcomes for inspection.
For exact fields, values and request limits, consult the HTTP API reference; do not infer the full contract from this guide.