← All examples 72 / 100

Jev Claim verification priority Code Example

Flag specific factual claims that need a source.

Our tool reduces costs by 73% for every company.

Sentence Our tool reduces costs by 73% for every company.

Source none

What editorial action is needed?

0% Request evidence.

Request evidence 86%

Ready to publish 10%

Copy edit 4%

Ask for what is needed before choosing request evidence.

Code

// Conceptual pseudocode — not Jev SDK syntax
const state = {
  "sentence": "Our tool reduces costs by 73% for every company.",
  "source": "none"
};

const decision = await evaluateDecision({
  state,
  question: "What editorial action is needed?",
  options: ["Request evidence","Ready to publish","Copy edit"],
});

// Your adapter normalizes the provider response.
// Tune this example threshold on labelled data.
if (decision.outcome === "Request evidence"
    && decision.probability >= 0.85) {
  flagUnverifiedClaim(decision);
} else {
  queueForReview(decision);
}

This adapter illustrates the decision boundary. Check the official Jev docs for current API syntax.