Notes

//This Analysis will combine tf/idf and term count and spanquery scores but apply adjusting weights from app setting
//I found an issue here. TFIDF is never used
private void btnCustAnal3_Click(object sender, EventArgs e)
{
//perform a rank search and store the results
TopDocs rankSearch = simpleRank();
TopDocs spanSearch = spanQuery();
TopDocs tfIdfSearch = tfIdfQuery();
if (rankSearch != null && spanSearch != null && tfIdfSearch != null)
{
directory = FSDirectory.Open(new DirectoryInfo(appDefaultindexPath));
IndexReader indexReader = IndexReader.Open(directory, true);
IndexSearcher indexSearch = new IndexSearcher(indexReader);
BooleanQuery booleanquery = new BooleanQuery();
String tmp = "";
String docText = "";
String strTypeSearch = "B Analysis";//"Rank and Span Query with IDF Tie Breakers";
var spanHit = spanSearch.ScoreDocs.OrderByDescending(o => o.Score).ToList();
var tfHit = tfIdfSearch.ScoreDocs.OrderByDescending(s => s.Score).ToList();
//I am not certain but I think this is wrong it is applying a tfidf score if there is no span score
//Tuesday this is not where the issue is.
//String outPut = "";
//foreach (ScoreDoc rd in rankSearch.ScoreDocs)
//{
// outPut += rd.Score.ToString() + ", ";
//}
//MessageBox.Show(outPut.ToString());
//Friday 3-2 I may need to add tf idf to each here regardless of spans object
float highestRank = rankSearch.MaxScore;
//MessageBox.Show("The highest rank is: " + highestRank.ToString());
float minRank = 0;
foreach (ScoreDoc rd in rankSearch.ScoreDocs)
{

var thsSd = spanHit.Find(o => o.Doc == rd.Doc);
var thsTf = tfHit.Find(s => s.Doc == rd.Doc);
//Normalize the TF and SPAN score so it never overwhelms the Rank of 1
//if (rd.Score > 2) { MessageBox.Show("Rank score is: " + rd.ToString()); }

//MessageBox.Show("The highes rank: " + rd.Score);
//if this is found there is a span hit document
if (thsSd != null)
{
rd.Score += Math.Abs((thsSd.Score - minRank) / (highestRank - minRank));
//rd.Score += thsSd.Score;
}
//2-9-17 removed the else from this. it is double counting
//else //there is no span to score just use the tf idf score
//{
//var thsTf = tfHit.Find(s => s.Doc == rd.Doc);
else if (thsTf != null)
{
//MessageBox.Show("No Span query");
//rd.Score = rd.Score + (thsTf.Score * tfRatio);
///MessageBox.Show("RdScore = " + rd.Score.ToString() + " new:" + (rd.Score + thsTf.Score).ToString());
rd.Score += Math.Abs((thsTf.Score - minRank) / (highestRank - minRank));
//rd.Score += thsTf.Score;
//var documentTest = indexSearch.Doc(rd.Doc);
//String sf = documentTest.Get("BODY");
//String[] phrs = sf.Split(new char[0], StringSplitOptions.RemoveEmptyEntries);
//sf
//indexSearch.Explain()
//phrs.Length();

//MessageBox.Show("Terms in Doc: " + rd.Doc.ToString());
}
//}
}
var hits = rankSearch.ScoreDocs.OrderByDescending(o => o.Score).ToList();
//just get the size we want
int sz = Convert.ToInt32(cmbBoxReturnSize.Text);
int cntr = 0;
foreach (var hit in hits)
{
if (cntr >= sz)
{
break;
}
var documentFromSearch = indexSearch.Doc(hit.Doc);
docText = documentFromSearch.Get("BODY");
tmp = documentFromSearch.Get("CCR");
//look for duplicate
var ths = scoredDocs.Where(c => c.ccr == tmp);
if (ths.Count() < 1)
{
scoredDocs.Add(new scrDoc(hit.Doc, tmp, hit.Score, docText));
}
//scoredDocs.Add(new scrDoc(hit.Doc, tmp, hit.Score, docText));
cntr++;
}
//now lets create the child form to display them
if (this.OwnedForms.Count() < 1)
{
intCntofChildren = 1;
}
else
{
intCntofChildren++;
}
Form srchResultForm = new SAT_Tool.ScoreForm(indexReader, indexSearch, booleanquery, scoredDocs, appDefaultindexPath, corpusDefaultPath, SrcTrms, SynTrms, phraseSlop, intCntofChildren, strTypeSearch, getQuery());
srchResultForm.Owner = this;
srchResultForm.Show();
}

}