public void AddTfIdfToDocuments()
{
foreach (var term in invertedIndex.Keys)
{
double idf = invertedIndex[term].Idf;
TermPostingList postingLine = GetPostingLineFromPostingFile(term, invertedIndex[term].PointerLine);
foreach (var termData in postingLine.PostingList)
{
DocumentsDataList[termData.DocumentId].TotalSquaredTfIdf += Math.Pow((termData.TF / DocumentsDataList[termData.DocumentId].FrequentTermNumberOfInstances) * idf, 2);
}
}
}
/// <summary>
/// Gets all the posting list of the terms in the query
/// </summary>
/// <param name="termsInQuery">List of terms in the query</param>
/// <returns>Dictionary of the term and it's posting list</returns>
private Dictionary <string, TermPostingList> GetQueryPostingList(List <string> termsInQuery)
{
Dictionary <string, TermPostingList> queryPostingsList = new Dictionary <string, TermPostingList>();
foreach (var term in termsInQuery) // for each term in the query
{
TermPostingList termPostingList = null;
if (_controller.InvertedIndex.ContainsKey(term)) // case the term appearce in the inverted index
{
//switch (_controller.InvertedIndex[term].TermDictionaryPointer) // check the location of the posting list
//{
//case TermDictionaryPointer.Cache: // case the posting line is in the cache, extract from the cacheDictionary
// termPostingList = _controller.CacheDictionary[term].PostingList;
// break;
//case TermDictionaryPointer.Posting:// case the posting line is in the posting file, extract from the file
termPostingList = GetPostingLineFromPostingFile(term, _controller.InvertedIndex[term].PointerLine);
//break;
// }
queryPostingsList.Add(term, termPostingList); // add the posting list to the dictionary
}
}
return(queryPostingsList);
}
