/**
* Set the original query terms.
*
* @param query The original query.
*/
public void setOriginalQueryTerms(MatchingQueryTerms query) {
String[] terms = query.getTerms();
this.originalTermids.clear();
for (int i = 0; i < terms.length; i++) {
EntryStatistics te = query.getStatistics(terms[i]);
if (te != null) {
this.originalTermids.put(te.getTermId(), terms[i]);
this.originalTermFreqs.adjustOrPutValue(
te.getTermId(), query.getTermWeight(terms[i]), query.getTermWeight(terms[i]));
}
}
}
/**
* This method implements the functionality of expanding a query.
*
* @param query MatchingQueryTerms the query terms of the original query.
* @param rq the Request thus far, giving access to the query and the result set
*/
public void expandQuery(MatchingQueryTerms query, Request rq) throws IOException {
// the number of term to re-weight (i.e. to do relevance feedback) is
// the maximum between the system setting and the actual query length.
// if the query length is larger than the system setting, it does not
// make sense to do relevance feedback for a portion of the query. Therefore,
// we re-weight the number of query length of terms.
int numberOfTermsToReweight = Math.max(ApplicationSetup.EXPANSION_TERMS, query.length());
if (ApplicationSetup.EXPANSION_TERMS == 0) numberOfTermsToReweight = 0;
if (selector == null) selector = this.getFeedbackSelector(rq);
if (selector == null) return;
FeedbackDocument[] feedback = selector.getFeedbackDocuments(rq);
if (feedback == null || feedback.length == 0) return;
// double totalDocumentLength = 0;
// for(FeedbackDocument doc : feedback)
// {
// totalDocumentLength += documentIndex.getDocumentLength(doc.docid);
// if(logger.isDebugEnabled()){
// logger.debug(doc.rank +": " + metaIndex.getItem("docno", doc.docid)+
// " ("+doc.docid+") with "+doc.score);
// }
// }
ExpansionTerms expansionTerms = getExpansionTerms();
expansionTerms.setModel(QEModel);
for (FeedbackDocument doc : feedback) {
expansionTerms.insertDocument(doc);
}
logger.debug(
"Selecting "
+ numberOfTermsToReweight
+ " from "
+ expansionTerms.getNumberOfUniqueTerms());
expansionTerms.setOriginalQueryTerms(query);
SingleTermQuery[] expandedTerms = expansionTerms.getExpandedTerms(numberOfTermsToReweight);
for (int i = 0; i < expandedTerms.length; i++) {
SingleTermQuery expandedTerm = expandedTerms[i];
query.addTermPropertyWeight(expandedTerm.getTerm(), expandedTerm.getWeight());
if (logger.isDebugEnabled()) {
logger.debug(
"term "
+ expandedTerms[i].getTerm()
+ " appears in expanded query with normalised weight: "
+ Rounding.toString(query.getTermWeight(expandedTerms[i].getTerm()), 4));
}
}
}
/*
* (non-Javadoc)
*
* @see org.irlib.features.Feature#computeValue()
*/
@Override
public Collection<? extends Double> computeValue() {
outputFeatures.removeAllElements();
double ctf = 1;
for (String term : terms.getTerms()) {
if (index.getLexicon().getLexiconEntry(term) != null)
ctf *=
(double) index.getLexicon().getLexiconEntry(term).getFrequency()
/ (double) totalNumberOfTokens;
}
outputFeatures.add((Math.log(ctf) / Math.log(2)) / terms.length());
return outputFeatures;
}
/**
* Runs the actual query expansion
*
* @see
* org.terrier.querying.PostProcess#process(org.terrier.querying.Manager,org.terrier.querying.SearchRequest)
*/
public void process(Manager manager, SearchRequest q) {
Index index = getIndex(manager);
lastIndex = index;
documentIndex = index.getDocumentIndex();
invertedIndex = index.getInvertedIndex();
lexicon = index.getLexicon();
collStats = index.getCollectionStatistics();
directIndex = index.getDirectIndex();
metaIndex = index.getMetaIndex();
if (directIndex == null) {
logger.error("This index does not have a direct index. Query expansion disabled!!");
return;
}
logger.debug("Starting query expansion post-processing.");
// get the query expansion model to use
String qeModel = q.getControl("qemodel");
if (qeModel == null || qeModel.length() == 0) {
logger.warn(
"qemodel control not set for QueryExpansion" + " post process. Using default model Bo1");
qeModel = "Bo1";
}
setQueryExpansionModel(getQueryExpansionModel(qeModel));
if (logger.isDebugEnabled()) {
logger.info("query expansion model: " + QEModel.getInfo());
}
MatchingQueryTerms queryTerms = ((Request) q).getMatchingQueryTerms();
if (queryTerms == null) {
logger.warn("No query terms for this query. Skipping QE");
return;
}
// get the expanded query terms
try {
expandQuery(queryTerms, (Request) q);
} catch (IOException ioe) {
logger.error("IOException while expanding query, skipping QE", ioe);
return;
}
if (logger.isDebugEnabled()) {
logger.info("query length after expansion: " + queryTerms.length());
logger.info("Expanded query: ");
}
final String[] newQueryTerms = queryTerms.getTerms();
StringBuilder newQuery = new StringBuilder();
for (int i = 0; i < newQueryTerms.length; i++) {
try {
if (logger.isDebugEnabled()) {
logger.info(
(i + 1)
+ ": "
+ newQueryTerms[i]
+ ", normalisedFrequency: "
+ Rounding.toString(queryTerms.getTermWeight(newQueryTerms[i]), 4));
}
newQuery.append(newQueryTerms[i]);
newQuery.append('^');
newQuery.append(Rounding.toString(queryTerms.getTermWeight(newQueryTerms[i]), 9));
newQuery.append(' ');
} catch (NullPointerException npe) {
logger.error("Nullpointer exception occured in Query Expansion dumping of new Query", npe);
}
}
logger.debug("NEWQUERY " + q.getQueryID() + " " + newQuery.toString());
lastExpandedQuery = newQuery.toString();
q.setControl("QE.ExpandedQuery", newQuery.toString());
final boolean no2ndPass =
Boolean.parseBoolean(ApplicationSetup.getProperty("qe.no.2nd.matching", "false"));
if (no2ndPass) {
return;
}
// run retrieval process again for the expanded query
logger.info("Accessing inverted file for expanded query " + q.getQueryID());
manager.runMatching(q);
}
