public double CalcRankFactor(WordOccurrenceNode occ, Query query)
{
double queryRank = 0.0;
int countTermQuery = 0;
foreach (QueryItem item in query.QueryItens)
{
if (item.WordID == occ.Word.WordID)
{
countTermQuery++;
}
}
int countTermDoc = occ.Hits.Count;
double bm25_TF = ((occ.Word.QuantityDocFrequency + 1)*countTermDoc)/(countTermDoc + occ.Word.QuantityDocFrequency);
double tf = occ.Frequency;
double idf = Math.Log((((double)totalDocQuantity) + 1) / ((double)occ.Word.QuantityDocFrequency));
double tf_idf = ((double)countTermQuery) * (bm25_TF) * idf;
queryRank += tf_idf;
return queryRank;
}
public double CalcRankFactor(WordOccurrenceNode occ, Query query)
{
double queryRank = 0.0;
int countTermQuery = 0;
foreach (QueryItem item in query.QueryItens)
{
if (item.WordID == occ.Word.WordID)
{
countTermQuery++;
}
}
double qtf = ((double)countTermQuery / (double)query.QueryItens.Count);
double termQueryFactor = ((k3 + 1) * qtf) / (k3 + qtf);
double df = (double)occ.Word.QuantityDocFrequency;
//double termLogFactor = Math.Log( ((totalDocQuantity - df + 0.5D)/(df + 0.5D)),Math.E);
double termLogFactor = Math.Log( ((double)totalDocQuantity) / ((double)df));
double tf = ((double)occ.Hits.Count / (double)occ.Doc.WordQuantity);
double normalizer = ((k1*(1 - b)) + (b * (occ.Doc.WordQuantity / avdl))) + tf;
double normalizationTermFactor = ((k1 + 1) * tf) / normalizer;
queryRank = termLogFactor * normalizationTermFactor * termQueryFactor;
return queryRank;
}
private List<Word> FindWords(Query parsedQuery)
{
List<Word> wordFound = new List<Word>();
for (int i = 0; ((i < parsedQuery.QueryItens.Count) && (i < maxSentence)); i++)
{
Word wf = indexer.Search(parsedQuery.QueryItens[i].WordID);
if (wf != null)
wordFound.Add(wf);
}
return wordFound;
}
public List<DocumentResult> Search(string query)
{
Hashtable resultHash = new Hashtable();
List<DocumentResult> resultList = new List<DocumentResult>();
Query parsedQuery = new Query(query);
List<Word> wordFound = FindWords(parsedQuery);
//merging the list.
foreach (Word item in wordFound)
{
List<WordOccurrenceNode> tempDocList = InvertedFileManager.Instance.GetWordOccurrencies(item);
foreach (WordOccurrenceNode wordOccur in tempDocList)
{
if (!resultHash.ContainsKey(wordOccur.Doc.DocID))
{
DocumentResult newDoc = new DocumentResult(wordOccur.Doc);
newDoc.CalculateRank(wordOccur, parsedQuery);
resultHash.Add(newDoc.DocID, newDoc);
}
else
{
DocumentResult newDoc = resultHash[wordOccur.Doc.DocID] as DocumentResult;
newDoc.CalculateRank(wordOccur, parsedQuery);
}
}
}
//convert hasthtable to list
foreach (DictionaryEntry entry in resultHash)
{
DocumentResult doc = entry.Value as DocumentResult;
resultList.Add(doc);
}
//sort result list by QueryRank and return
resultList.Sort((y, x) => x.QueryRank.CompareTo(y.QueryRank));
return resultList;
}
public void CalculateRank(WordOccurrenceNode occ, Query query)
{
IRankFunction rankFunc = FactoryRankFunction.GetRankFunction();
this.queryRank += rankFunc.CalcRankFactor(occ, query);
}
public List<DocumentResult> Search(string query)
{
Hashtable resultHash = new Hashtable();
List<DocumentResult> resultList = new List<DocumentResult>();
Query parsedQuery = new Query(query);
List<Word> wordFound = FindWords(parsedQuery);
//merging the list.
foreach (Word item in wordFound)
{
WordOccurrenceNode firstOcc = item.FirstOccurrence;
//problem: the number of occurrences is wrong! The 'else' case, doesn't exist and because this,
//the program don't count the occurrences of the second word.
//when he merge, it discards the occurrences.
if (!resultHash.ContainsKey(firstOcc.Doc.DocID))
{
DocumentResult newDoc = new DocumentResult(firstOcc.Doc);
newDoc.CalculateRank(firstOcc, parsedQuery);
resultHash.Add(newDoc.DocID, newDoc);
}
else
{
DocumentResult newDoc = resultHash[firstOcc.Doc.DocID] as DocumentResult;
newDoc.CalculateRank(firstOcc, parsedQuery);
}
WordOccurrenceNode tmp = firstOcc;
while (tmp.HasNext())
{
tmp = tmp.NextOccurrence;
if (!resultHash.ContainsKey(tmp.Doc.DocID))
{
DocumentResult newDoc = new DocumentResult(tmp.Doc);
newDoc.CalculateRank(tmp, parsedQuery);
resultHash.Add(newDoc.DocID, newDoc);
}
else
{
DocumentResult newDoc = resultHash[tmp.Doc.DocID] as DocumentResult;
newDoc.CalculateRank(tmp, parsedQuery);
}
}
}
//convert hasthtable to list
foreach (DictionaryEntry entry in resultHash)
{
DocumentResult doc = entry.Value as DocumentResult;
resultList.Add(doc);
}
//sort result list by QueryRank and return
resultList.Sort((y, x) => x.QueryRank.CompareTo(y.QueryRank));
return resultList;
}
