mongodb-rag-core/build/executeCode/computeNormalizedLogarithmicQueryEfficiency.js

Common elements used by MongoDB Chatbot Framework components.

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.computeNormalizedLogarithmicQueryEfficiency = void 0;
const assert_1 = require("assert");
/**
  Computes a normalized query efficiency score based on the logarithmic relationship
  between documents examined and documents returned. The score ranges from 0 to 1, where:
 
  - 1 means perfect efficiency (n_examined = n_returned)
  - 0 means full scan with minimal results
  - Score decreases logarithmically as examination overhead increases
  - Formula: 1 - log((n_examined+eps2)/(n_returned+eps2)) / log((n_total+eps1)/(n_returned+eps2))
  - Uses epsilon values for numerical stability and to handle zero-result queries

  @returns A score between 0 and 1
 */
function computeNormalizedLogarithmicQueryEfficiency({ nReturned, nExamined, nTotal, }) {
    (0, assert_1.strict)(nReturned >= 0 && nExamined >= 0 && nTotal > 0, "Document counts must be non-negative, and total must be > 0");
    (0, assert_1.strict)(nExamined >= nReturned, "Documents examined must be >= documents returned");
    (0, assert_1.strict)(nTotal >= nExamined, "Total documents must be >= documents examined");
    // Epsilon values for numerical stability
    const eps1 = 0.1; // Added to nTotal
    const eps2 = 0.01; // Added to nReturned and nExamined
    // Formula with epsilon smoothing
    const efficiency = 1 -
        Math.log((nExamined + eps2) / (nReturned + eps2)) /
            Math.log((nTotal + eps1) / (nReturned + eps2));
    // Clamp between 0 and 1
    return Math.max(0, Math.min(1, efficiency));
}
exports.computeNormalizedLogarithmicQueryEfficiency = computeNormalizedLogarithmicQueryEfficiency;
//# sourceMappingURL=computeNormalizedLogarithmicQueryEfficiency.js.map