/**
* Extracts termClaimsDescriptionAbstractTitles of the documents; Adds them to vector in the same
* order
*
* @param hits
* @param i
* @param j
* @return relevantDocsTerms docs must be in order
* @throws java.io.IOException
*/
public Map<TermFreqVector, String> getDocsTerms(TopDocs hits, int i, int j) throws IOException {
Map<TermFreqVector, String> docsTerms = new HashMap<>();
// Process each of the documents
while (i < j && i < hits.totalHits && i >= 0) {
ScoreDoc scoreDoc = hits.scoreDocs[i];
if (sourceField.equals(PatentQuery.all)) {
Terms termTitle = ir.getTermVector(scoreDoc.doc, PatentQuery.getFields()[1]);
TermFreqVector docTermsTitle = new TermFreqVector(termTitle);
docsTerms.put(docTermsTitle, PatentQuery.getFields()[1]);
Terms termAbstract = ir.getTermVector(scoreDoc.doc, PatentQuery.getFields()[2]);
TermFreqVector docTermsAbstract = new TermFreqVector(termAbstract);
docsTerms.put(docTermsAbstract, PatentQuery.getFields()[2]);
Terms termDescription = ir.getTermVector(scoreDoc.doc, PatentQuery.getFields()[3]);
TermFreqVector docTermsDescription = new TermFreqVector(termDescription);
docsTerms.put(docTermsDescription, PatentQuery.getFields()[3]);
Terms termClaims = ir.getTermVector(scoreDoc.doc, PatentQuery.getFields()[5]);
TermFreqVector docTermsClaims = new TermFreqVector(termClaims);
docsTerms.put(docTermsClaims, PatentQuery.getFields()[5]);
} else {
Terms term = ir.getTermVector(scoreDoc.doc, sourceField); // get termvector for document
// Create termVector and add it to vector
TermFreqVector docTerms = new TermFreqVector(term);
docsTerms.put(docTerms, sourceField);
}
i++;
}
return docsTerms;
}
/**
* Creates a new instance of QueryExpansion
*
* @param hits
* @param ir
* @param parameters
* @param source
* @param Nbr_Docs
* @param Nbr_Terms
* @throws java.io.IOException
*/
public RocchioQueryExpansion(
TopDocs hits,
IndexReader ir,
Map<String, Float> parameters,
int source,
int Nbr_Docs,
int Nbr_Terms)
throws IOException {
this.ir = ir;
this.parameters = parameters;
if (source != 7) {
this.sourceField = PatentQuery.getFields()[source];
} else {
this.sourceField = PatentQuery.all;
}
this.Nbr_Terms = Nbr_Terms;
// Create combine documents termTitle vectors - sum ( rel termTitle vectors )
// Get terms from relevant documents
docsTermVectorReldocs = getDocsTerms(hits, 0, Nbr_Docs);
// Get terms from irrelevant documents
docsTermVectorIrreldocs = getDocsTerms(hits, hits.totalHits - Nbr_Docs, hits.totalHits);
}