/**
* Adjust termClaimsDescriptionAbstractTitle features of the docs with alpha * query; and beta;
* and assign weights/boost to termClaimsDescriptionAbstractTitles (tf*idf).
*
* @param query
* @param currentField
* @param alpha
* @param beta - factor of the equation
* @param gamma
* @param decay
* @param maxExpandedQueryTerms - maximum number of termClaimsDescriptionAbstractTitles in
* expanded query
* @return expandedQuery with boost factors adjusted using Rocchio's algorithm
* @throws IOException
* @throws ParseException
*/
public Query adjust(
Query query,
String currentField,
float alpha,
float beta,
float gamma,
float decay,
int maxExpandedQueryTerms)
throws IOException, ParseException {
Query expandedQuery;
// setBoost of docs terms
Map<String, TermQuery> relevantDocsTerms =
setBoost(docsTermVectorReldocs, currentField, beta, decay);
Map<String, TermQuery> irrrelevantDocsTerms =
setBoost(docsTermVectorIrreldocs, currentField, gamma, decay);
// Map<String, TermQuery> relevantDocsTerms = new HashMap<>();
// Map<String, TermQuery> irrrelevantDocsTerms = new HashMap<>();
// setBoost of query terms
// Get queryTerms from the query
// combine weights according to expansion formula
List<TermQuery> expandedQueryTerms =
combine(new HashMap<String, TermQuery>(), relevantDocsTerms, irrrelevantDocsTerms);
// Sort by boost=weight
Comparator comparator = new QueryBoostComparator();
Collections.sort(expandedQueryTerms, comparator);
relevantDocsTerms.clear();
int termCount = Math.min(expandedQueryTerms.size(), maxExpandedQueryTerms);
for (int i = 0; i < termCount; i++) {
TermQuery tq = expandedQueryTerms.get(i);
relevantDocsTerms.put(tq.getTerm().text(), tq);
System.out.print(tq.getTerm().text() + ", ");
}
TermFreqVector queryTermsVector = new TermFreqVector(query);
Map<String, TermQuery> queryTerms;
queryTerms = setBoost(queryTermsVector, currentField, alpha);
// List<TermQuery> queryTermsList=new ArrayList(queryTerms.values());
// Collections.sort(queryTermsList, comparator);
// queryTerms.clear();
// for(TermQuery tq:queryTermsList){
// queryTerms.put(tq.getTerm().text(), tq);
// }
expandedQueryTerms = combine(queryTerms, relevantDocsTerms, new HashMap<String, TermQuery>());
Collections.sort(expandedQueryTerms, comparator);
// Create Expanded Query
expandedQuery = mergeQueries(expandedQueryTerms, Integer.MAX_VALUE);
return expandedQuery;
}
public Map<String, Float> getRocchioVector(String currentField) throws IOException {
Map<String, Float> out = new HashMap<>();
float beta = parameters.get(RocchioQueryExpansion.ROCCHIO_BETA_FLD);
float gamma = parameters.get(RocchioQueryExpansion.ROCCHIO_GAMMA_FLD);
float decay = parameters.get(RocchioQueryExpansion.DECAY_FLD);
Map<String, TermQuery> relevantDocsTerms =
setBoost(docsTermVectorReldocs, currentField, beta, decay);
Map<String, TermQuery> irrrelevantDocsTerms =
setBoost(docsTermVectorIrreldocs, currentField, gamma, decay);
List<TermQuery> expandedQueryTerms =
combine(
new HashMap<String, TermQuery>(), relevantDocsTerms, new HashMap<String, TermQuery>());
for (TermQuery tq : expandedQueryTerms) {
out.put(tq.getTerm().text(), tq.getBoost());
}
return out;
}
