public Dictionary <string, double> ProcessQueryTemp(string query)
{
var termsToCompute = GetTerms(query);
var termsQueryFrequency = GetTermsFrequency(termsToCompute);
var nrTermsToCompute = termsToCompute.Count;
var termsIndex = IndexRepo.GetIndexOfWords(termsToCompute);
var documentsLength = GetDocumentsLength(termsIndex);
var documentsScoresAccumulator = new Dictionary <string, double>();
foreach (var termIndex in termsIndex)
{
var term = termIndex.Term;
var docs = termIndex.Docs;
var nrDocs = docs.Count;
foreach (var doc in docs)
{
var tf = ComputeTermFrequency(doc.Frequency, documentsLength[doc.ID]);
var idf = ComputeInverseDocumentFrequency(nrDocs);
if (!documentsScoresAccumulator.ContainsKey(doc.ID))
{
documentsScoresAccumulator.Add(doc.ID, 0);
}
documentsScoresAccumulator[doc.ID] += (((double)termsQueryFrequency[term]) / nrTermsToCompute) * tf * idf;
}
}
return(documentsScoresAccumulator);
}
public Dictionary <string, double> ProcessQueryTemp(string query)
{
var termsToCompute = GetTerms(query);
var termsIndex = IndexRepo.GetIndexOfWords(termsToCompute);
var termsQueryPositions = GetTermsPosition(termsToCompute);
var termsWeight = GetTermsWeight(termsIndex);
var documentsToScore = ComputeDocumentsIndexFromTermsIndex(termsIndex);
var documentsLen = GetDocumentsLength(termsIndex);
var result = new Dictionary <string, double>();
foreach (var docToScore in documentsToScore)
{
var docID = docToScore.Key;
var fti = new Dictionary <string, double>();
var docLen = documentsLen[docID];
var termsRelativePositions = CustomMergeSort.MergeKSortedLists(docToScore.Value);
var docScore = 0.0;
for (int i = 0; i < termsRelativePositions.Count; ++i)
{
var termPos = termsRelativePositions[i].Key;
var term = termsRelativePositions[i].Value;
var minDistanceElementIndex = ComputeMinDistanceElementIndex(i, termsRelativePositions);
var documentOffset = termPos - termsRelativePositions[minDistanceElementIndex].Key;
var queryOffset = termsQueryPositions[term] - termsQueryPositions[termsRelativePositions[minDistanceElementIndex].Value];
if (!fti.ContainsKey(term))
{
fti.Add(term, DefaultTermScore);
}
var dist = Math.Abs(documentOffset - queryOffset);
fti[term] += termsWeight[term] * termsWeight[termsRelativePositions[minDistanceElementIndex].Value] * ReverseProximity(dist);
}
var K = ComputeK(docLen);
foreach (var item in fti)
{
docScore += termsWeight[item.Key] * item.Value / (K + item.Value);
}
result.Add(docID, docScore);
}
return(result);
}
public List <string> AND(List <string> termsToCompute)
{
var termsIndex = IndexRepo.GetIndexOfWords(termsToCompute);
var preResult = new HashSet <string>();
if (termsIndex.Count == 0)
{
return(preResult.ToList());
}
preResult.UnionWith(termsIndex[0].Docs.Select(x => x.ID));
foreach (var item in termsIndex.Skip(1))
{
preResult.IntersectWith(item.Docs.Select(x => x.ID));
}
return(preResult.ToList());
}
public List <string> OR(List <string> termsToCompute)
{
var termsIndex = IndexRepo.GetIndexOfWords(termsToCompute);
return(ComputeOR(termsIndex));
}