wordmap/dist/algorithms/AlignmentOccurrences.js

Multi-Lingual Word Alignment Prediction

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
const Algorithm_1 = require("../Algorithm");
/**
 * A commonly seen pattern in translation is that word repetition in the primary text
 * is often seen in the secondary text.
 */
class AlignmentOccurrences extends Algorithm_1.default {
    constructor() {
        super(...arguments);
        this.name = "alignment occurrences";
    }
    /**
     * Calculates the
     * @param sourceNgramFrequency
     * @param targetNgramFrequency
     */
    static calc(sourceNgramFrequency, targetNgramFrequency) {
        if (sourceNgramFrequency === 0 || targetNgramFrequency === 0) {
            return 0;
        }
        else {
            return Math.min(sourceNgramFrequency, targetNgramFrequency) /
                Math.max(sourceNgramFrequency, targetNgramFrequency);
            // TODO: Review above change with Klappy
            // const delta = Math.abs(sourceNgramFrequency - targetNgramFrequency);
            // return 1 / (delta + 1);
        }
    }
    /**
     * Calculates the weight based on the word
     * @param p
     * @param usIndex
     */
    static calcOccurrenceSimilarity(p, usIndex) {
        const sourceFrequency = usIndex.static.sourceNgramFrequency.read(p.source);
        const targetFrequency = usIndex.static.targetNgramFrequency.read(p.target);
        const weight = AlignmentOccurrences.calc(sourceFrequency, targetFrequency);
        p.setScore("alignmentOccurrences", weight);
    }
    /**
     * Calculates the weight based on the lemma
     * @param p
     * @param usIndex
     */
    static calcLemmaOccurrenceSimilarity(p, usIndex) {
        if (p.source.lemmaKey !== undefined && p.target.lemmaKey !== undefined) {
            const sourceFrequency = usIndex.static.sourceNgramFrequency.read(p.source.lemmaKey);
            const targetFrequency = usIndex.static.targetNgramFrequency.read(p.target.lemmaKey);
            const weight = AlignmentOccurrences.calc(sourceFrequency, targetFrequency);
            p.setScore("lemmaAlignmentOccurrences", weight);
        }
        else {
            p.setScore("lemmaAlignmentOccurrences", 0);
        }
    }
    execute(prediction, cIndex, saIndex, usIndex) {
        AlignmentOccurrences.calcOccurrenceSimilarity(prediction, usIndex);
        AlignmentOccurrences.calcLemmaOccurrenceSimilarity(prediction, usIndex);
        return prediction;
    }
}
exports.default = AlignmentOccurrences;