public void initTonNDocs(Set<String> docs_topn) {
if (docs_topn == _mPrevInit) return; // Already initialized
_mPrevInit = docs_topn;
_hmKey2IDF = new HashMap<Object, Double>();
double total_doc_length = 0d;
// Have to go through all documents to compute term IDF
for (String doc : docs_topn) {
// Map<Object,Double> features = (Map<Object,Double>)getObjectRepresentation(doc);
Map<Object, Double> features = DocUtils.ConvertToFeatureMap(_docs_all.get(doc));
//
total_doc_length += VectorUtils.L1Norm(features);
//
features = VectorUtils.ConvertToBoolean(features);
//
_hmKey2IDF = VectorUtils.Sum(_hmKey2IDF, features);
/*
if (false && ++i % 100 == 0) {
System.out.println("- Converted " + i + " documents to internal representation");
System.out.flush();
}
*/
}
_dAvgDocLength = total_doc_length / (double) docs_topn.size();
for (Object key : _hmKey2IDF.keySet()) {
Double idf = docs_topn.size() / (_hmKey2IDF.get(key) + 1d);
_hmKey2IDF.put(key, Math.log(idf));
}
_dDefaultIDF = Math.log(docs_topn.size() / 1d);
if (DEBUG) {
System.out.println("Avg doc length: " + _dAvgDocLength);
System.out.println("Default IDF : " + _dDefaultIDF);
System.out.println("IDF after log: " + _hmKey2IDF);
}
}
public Object getNoncachedObjectRepresentation(String content) {
Map<Object, Double> features = DocUtils.ConvertToFeatureMap(content);
return features;
}
