private void calculateDocClassTermIDFs() {
docClassTermIDFs = new HashMap<>();
for (String term : docClassFrequencyPerTerm.getTerms()) {
Integer docFrequency = docClassFrequencyPerTerm.getFrequency(term);
Double docFraction = docClasses.size() / (1 + docFrequency.doubleValue());
docClassTermIDFs.put(term, Math.log10(docFraction));
}
}
private Map<String, Double> weightBagOfWordsWithIDF(BagOfWords queryBag) {
Map<String, Double> weighted = new HashMap<>();
for (Entry<String, Integer> entry : queryBag.getFrequencies().entrySet()) {
String term = entry.getKey();
Double classifierCorpusIDF = docClassTermIDFs.get(term);
weighted.put(
entry.getKey(),
(null == classifierCorpusIDF) ? 0d : classifierCorpusIDF * queryBag.getFrequency(term));
}
return weighted;
}