public HMM(int n, List<Pair> pairs) {
long time = System.currentTimeMillis();
this.n = n;
this.pairs = pairs;
System.out.println("Counting states");
// Find States and Labels
for (Pair pair : pairs) {
words.add(pair.word);
tags.add(pair.tag);
}
System.out.println("Finding transition probabilities");
System.out.println(pairs.size());
// Find tag transition probabilities
System.out.println("Counting ngrams");
countGrams(n, nGramCount, false);
if (n > 1) {
System.out.println("Counting n-1 grams");
countGrams(n - 1, nMinusOneGramCount, false);
}
// Find emission probabilities
System.out.println("Finding emission probabilities");
for (Pair pair : pairs) {
Util.tagWordCount(tagsMap, pair.word, pair.tag);
Util.wordTagCount(wordsMap, pair.word, pair.tag);
Util.incrementMap(wordCount, pair.word);
Util.incrementMap(tagCount, pair.tag);
}
Util.mostCommonTagMap(wordsMap, mostCommonTagMap);
mostCommonTag = Util.mostCommonTag(tagCount);
System.out.println((System.currentTimeMillis() - time) / 1000.0);
}
public void countGrams(int n, HashMap<String, Integer> count, boolean word) {
List<String> nGram = new ArrayList<String>();
int i;
for (i = 0; i < n; i++) nGram.add(pairs.get(i).getContent(word));
for (; i < pairs.size(); i++) {
Util.incrementMap(count, nGram.toString());
nGram.add(pairs.get(i).getContent(word));
nGram.remove(0);
}
Util.incrementMap(count, nGram.toString());
}
