/** Accumulate statistics on how often each token value occurs */
public void train(StringWrapperIterator i) {
Set seenTokens = new HashSet();
while (i.hasNext()) {
BagOfTokens bag = asBagOfTokens(i.nextStringWrapper());
seenTokens.clear();
for (Iterator j = bag.tokenIterator(); j.hasNext(); ) {
totalTokenCount++;
Token tokj = (Token) j.next();
if (!seenTokens.contains(tokj)) {
seenTokens.add(tokj);
// increment documentFrequency counts
Integer df = (Integer) documentFrequency.get(tokj);
if (df == null) documentFrequency.put(tokj, ONE);
else if (df == ONE) documentFrequency.put(tokj, TWO);
else if (df == TWO) documentFrequency.put(tokj, THREE);
else documentFrequency.put(tokj, new Integer(df.intValue() + 1));
}
}
collectionSize++;
}
}
public Iterator tokenIterator() {
return documentFrequency.keySet().iterator();
}
public int getDocumentFrequency(Token tok) {
Integer freqInteger = (Integer) documentFrequency.get(tok);
if (freqInteger == null) return 0;
else return freqInteger.intValue();
}