compromise/src/terms/tagger/steps/06-neighbour_step.js

natural language processing in the browser

'use strict';
const markov = require('./data/neighbours');
const afterThisWord = markov.afterThisWord;
const beforeThisWord = markov.beforeThisWord;
const beforeThisPos = markov.beforeThisPos;
const afterThisPos = markov.afterThisPos;
const log = require('../paths').log;
const path = 'tagger/neighbours';

//basically a last-ditch effort before everything falls back to a noun
//for unknown terms, look left + right first, and hit-up the markov-chain for clues
const neighbour_step = function (ts) {
  log.here(path);
  ts.terms.forEach((t, n) => {
    //is it still unknown?
    let termTags = Object.keys(t.tag);
    if (termTags.length === 0) {
      let lastTerm = ts.terms[n - 1];
      let nextTerm = ts.terms[n + 1];
      //look at last word for clues
      if (lastTerm && afterThisWord[lastTerm.normal]) {
        t.tagAs(afterThisWord[lastTerm.normal], 'neighbour-after-"' + lastTerm.normal + '"');
        return;
      }
      //look at next word for clues
      if (nextTerm && beforeThisWord[nextTerm.normal]) {
        t.tagAs(beforeThisWord[nextTerm.normal], 'neighbour-before-"' + nextTerm.normal + '"');
        return;
      }
      //look at the last POS for clues
      let tags = [];
      if (lastTerm) {
        tags = Object.keys(lastTerm.tag);
        for (let i = 0; i < tags.length; i++) {
          if (afterThisPos[tags[i]]) {
            t.tagAs(afterThisPos[tags[i]], 'neighbour-after-[' + tags[i] + ']');
            return;
          }
        }
      }
      //look at the next POS for clues
      if (nextTerm) {
        tags = Object.keys(nextTerm.tag);
        for (let i = 0; i < tags.length; i++) {
          if (beforeThisPos[tags[i]]) {
            t.tagAs(beforeThisPos[tags[i]], 'neighbour-before-[' + tags[i] + ']');
            return;
          }
        }
      }
    }
  });

  return ts;
};

module.exports = neighbour_step;