/** Compute number of correct pairs betwn src1 and src2, where src2>src1 */
protected int countCorrectPairs(MatchData data) {
// count the number of times each id appears in each source */
Map counter = new HashMap();
for (int i = 0; i < data.numSources(); i++) {
String src = data.getSource(i);
for (int j = 0; j < data.numInstances(src); j++) {
String id = data.getInstance(src, j).getId();
if (id != null) {
IdKey key = new IdKey(id, src);
Integer c = (Integer) counter.get(key);
counter.put(key, (c == null ? new Integer(1) : new Integer(c.intValue() + 1)));
}
}
}
/*
// show the counter
for (Iterator i=counter.keySet().iterator(); i.hasNext(); ) {
IdKey key = (IdKey) i.next();
System.out.println( key.src+"#"+key.id+" = "+counter.get(key) );
}
*/
// count the number of correct pairs
int numCorrectPairs = 0;
Set idsInSrc1 = new HashSet();
for (int i = 0; i < data.numSources(); i++) {
String src1 = data.getSource(i);
idsInSrc1.clear();
for (int j = 0; j < data.numInstances(src1); j++) {
String id = data.getInstance(src1, j).getId();
idsInSrc1.add(id);
for (int k = i + 1; k < data.numSources(); k++) {
String src2 = data.getSource(k);
Integer cInteger = (Integer) counter.get(new IdKey(id, src2));
if (cInteger != null) {
numCorrectPairs += cInteger.intValue();
}
// System.out.println( "src1:"+src1+" id:"+id+" src2:"+src2+" c:"+cInteger);
}
}
if (clusterMode) {
// count how often something in src1 can be matched correctly with something
// else in src1
for (Iterator j = idsInSrc1.iterator(); j.hasNext(); ) {
String id = (String) j.next();
Integer cInteger = (Integer) counter.get(new IdKey(id, src1));
int c = cInteger.intValue();
numCorrectPairs += c * (c - 1) / 2;
}
}
}
return numCorrectPairs;
}
/** 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++;
}
}