J.

Jev Review usefulness Code Example

Rank reviews that answer the shopper’s actual question.

Is it good for long-distance running?

Query Is it good for long-distance running?
Review Comfortable after 20 km; heel held firm.

How useful is this review for the query?

0% Highly useful.
Highly useful 78%
Somewhat useful 15%
Not useful 7%

Let the application highly useful.

Code
Conceptual pseudocode — not official Jev SDK syntax
// Conceptual pseudocode — not Jev SDK syntax
const state = {
  "query": "Is it good for long-distance running?",
  "review": "Comfortable after 20 km; heel held firm."
};

const decision = await evaluateDecision({
  state,
  question: "How useful is this review for the query?",
  options: ["Highly useful","Somewhat useful","Not useful"],
});

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

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