//Computes the TF-IDF rank of both the query and the documents relevant
//to the document
private void CompTF_IDF(Tuple <int, List <ITermDocument> > query)
{
//A cache of Length of each documents needed for ranking.
IDictionary <String, long> cacheDocumentsLength = new Dictionary <String, long>();
//No of document in Corpus.
long noDocs = _corpus.GetNoDocuments().Result;
//Compute the size of the query to compute each query's IDF
var querySize = query.Item1;
foreach (var term in query.Item2)
{
IDictionary <String, int> documents = new Dictionary <String, int>();
foreach (var doc in _corpus.GetDocuments(term.Term).Result)
{
if (documents.Keys.Contains(doc.DocID))
{
continue;
}
documents.Add(doc.DocID, doc.Pos.Count);
this._documents.Add(doc.DocID);
}
if (documents.Count < 1)
{
continue;
}
var nDocuments = new Dictionary <String, double>();
double IDF = (1 + Math.Log(1.0 * noDocs / documents.Keys.Count));
this._queryRank[term.Term] = (1.0 * term.Docs.First().Pos.Count / querySize) * IDF;
foreach (var item in documents)
{
long length;
if (cacheDocumentsLength.ContainsKey(item.Key))
{
length = cacheDocumentsLength[item.Key];
}
else
{
length = _corpus.GetDocumentLength(item.Key).Result;
cacheDocumentsLength[item.Key] = length;
}
if (length == 0)
{
continue;
}
double tF = 1.0 * item.Value / length;
nDocuments[item.Key] = tF * IDF;
}
_documentRank[term.Term] = nDocuments;
;
}
}
