double product(TokenizedCharSequence thatDoc) {
double sum = 0.0;
for (String token : mTokenCounter.keySet()) {
int count = thatDoc.mTokenCounter.getCount(token);
if (count == 0) continue;
// tf = sqrt(count); sum += tf1 * tf2
sum += Math.sqrt(count * mTokenCounter.getCount(token));
}
return sum;
}
static double length(ObjectToCounterMap<String> otc) {
double sum = 0.0;
for (Counter counter : otc.values()) {
double count = counter.doubleValue();
sum += count; // tf =sqrt(count); sum += tf * tf
}
return Math.sqrt(sum);
}
public TokenizedCharSequence(String input) throws IOException {
mString = input;
mText = input.toCharArray();
Tokenizer tokenizer = TOKENIZER_FACTORY.tokenizer(mText, 0, mText.length);
String token;
while ((token = tokenizer.nextToken()) != null) mTokenCounter.increment(token.toLowerCase());
mLength = length(mTokenCounter);
}
