Notes

/**
* This search scorer just counts the number of terms that are found in each document. It could leave out reports with too small of
* a return size selected by the user. This score has no value for relevance between documents as it relates to length or frequency
* of recurring terms.
* **/
public TopDocs simpleRank()
{
TopDocs EmptyDocs = new TopDocs(0, null, 1);
TopDocs rankDocs = null;
try
{
if (SrcTrms.Count > 0 && directory != null)//make sure there is an idea term to search with or else I cannot score documents
{
scoredDocs.Clear();
ScoreDoc[] ttlScoresArray = null;
String strTypeSearch = "Custom Rank Scoring";
BooleanQuery booleanquery = new BooleanQuery();
PhraseQuery phrasequery; //used for multi term queries
Query query; //used for single term queries
directory = FSDirectory.Open(new DirectoryInfo(appDefaultindexPath));
IndexReader indexReader = IndexReader.Open(directory, true);
IndexSearcher indexSearch = new IndexSearcher(indexReader);
//indexSearch.Similarity = DefaultSimilarity;
indexSearch.Similarity = new customDefaultScore();
//indexSearch.Similarity = new CustomScore();
var queryParser = new QueryParser(Version.LUCENE_30, "BODY", analyzer);
//get all the terms and boosts
foreach (SrcTrm i in SrcTrms)
{
float bst = 1;
//MessageBox.Show(i.boost);
String Boost = i.boost;
//determine if this is a phrase or single term
//if (i.idea.Contains(" "))
//{
//MessageBox.Show("This is a phrase");
//break the string into an array
phrasequery = new PhraseQuery();
String[] phrs = i.idea.Split(new char[0], StringSplitOptions.RemoveEmptyEntries);
int ctr = 0;
foreach (String str in phrs)
{
phrasequery.Add(new Term("BODY", str.ToLower()), ctr++);
//MessageBox.Show("The term is now: " + str);
}
phraseSlop = Convert.ToInt32(txtPhrSlop.Text);// Convert.ToInt32(cmbPhraseSlop.SelectedItem);
phrasequery.Slop = phraseSlop;
if (Boost == "Must")
{
bst = (float)Convert.ToInt32(Must_Weight);
//bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.MUST);
}
else if (Boost == "Never")
{
bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.MUST_NOT);
}
else if (Boost == "Probable")
{
bst = (float)Convert.ToInt32(Probable_Weight);
//bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
//booleanquery.Add(csQuery, Occur.SHOULD);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.SHOULD);
}
else if (Boost == "Possible")
{
bst = (float)Convert.ToInt32(Possible_Weight);
//bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
//booleanquery.Add(csQuery, Occur.SHOULD);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.SHOULD);
}
else if (Boost == "Critical")
{
bst = (float)Convert.ToInt32(Critical_Weight);
//bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
//booleanquery.Add(csQuery, Occur.SHOULD);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.SHOULD);
}
else
{
bst = 1;
//cstmScoreQry csQuery = new cstmScoreQry(phrasequery, bst, ttlScoresArray);
//booleanquery.Add(csQuery, Occur.SHOULD);
cstmRankQry rkQry = new cstmRankQry(phrs, phrasequery, bst, ttlScoresArray);
booleanquery.Add(rkQry, Occur.SHOULD);
}


}
rankDocs = indexSearch.Search(booleanquery, rankMaxSize);//indexReader.MaxDoc);
//foreach (ScoreDoc rd in rankSearch.ScoreDocs)
// foreach (ScoreDoc rd in rankDocs.ScoreDocs)
// {
// if(rd.Score > 2)
// {
// MessageBox.Show("Rank Score here: " + rd.Score.ToString());
// }
//}
return rankDocs;
}
}
return rankDocs;
}
/************************************************************************************
* This method goes through the ideas collection and provides the boost and terms
* to be recorded for the analysis phase.
* */
private String getQuery()
{
String theseTrms = "";
foreach (SrcTrm st in SrcTrms)
{
if (st.idea.Contains(" "))
{
//its a multi term phrase
theseTrms += "Phrase: " + st.idea.ToString() + " Boost: " + getBoostValue(st.boost.ToString()) + "rn";
}
else
{
theseTrms += "Term: " + st.idea.ToString() + " Boost: " + getBoostValue(st.boost.ToString()) + "rn";
}
}
return theseTrms;
}

/********************************************************************************
* This method uses a case select to determine the numerical boost values
* for the getQuery method.
*/


private TopDocs spanQuery()
{
TopDocs spResults = null;
try
{

//SpanQuery spnQry = new SpanNearQuery(new SpanQuery[] { new SpanTermQuery(new Term("BODY", "arrest")), new SpanTermQuery(new Term("BODY", "mirandized")) }, 30, false);
BooleanQuery booleanquery = new BooleanQuery();
PhraseQuery phrasequery; //used for multi term queries
Query query; //used for single term queries
directory = FSDirectory.Open(new DirectoryInfo(appDefaultindexPath));
IndexReader indexReader = IndexReader.Open(directory, true);
IndexSearcher indexSearch = new IndexSearcher(indexReader);
//indexSearch.Similarity = DefaultSimilarity;
var queryParser = new QueryParser(Version.LUCENE_30, "BODY", analyzer);
//TopDocs resultDocs = null;
//loop thru all the phrases
SpanQuery spNrQry = null;
List<SpanQuery> spnQry = new List<SpanQuery> { };
int slp = Convert.ToInt32(txtPhrSlop.Text);
foreach (SrcTrm i in SrcTrms)
{
//get the boost right off the bat
float bst = 1;
//MessageBox.Show("Idea is: " + i.idea );
if (i.idea.Contains(" ")) //if the idea is a phrase be careful of hyphens!!!
{
//MessageBox.Show("This is a phrase: " + i.idea);
String[] phrs = i.idea.Split(new char[0], StringSplitOptions.RemoveEmptyEntries);
foreach (String str in phrs)
{
//MessageBox.Show("str is: " + str);
spnQry.Add(new SpanTermQuery(new Term("BODY", str)));
}

spNrQry = new SpanNearQuery(spnQry.ToArray(), slp, false);
//MessageBox.Show("The span terms are: " + spNrQry.ToString());
//MessageBox.Show("This many spans: " + spResults.MaxScore );
//add this span query to the boolean query with the correct weights
if (i.boost == "Must")
{
bst = (float)Convert.ToInt32(Must_Weight);
spNrQry.Boost = bst;
booleanquery.Add(spNrQry, Occur.MUST);
}
else if (i.boost == "Never")
{
booleanquery.Add(spNrQry, Occur.MUST_NOT);
}
else if (i.boost == "Probable")
{
bst = (float)Convert.ToInt32(Probable_Weight);
spNrQry.Boost = bst;
booleanquery.Add(spNrQry, Occur.SHOULD);
}
else if (i.boost == "Possible")
{
bst = (float)Convert.ToInt32(Possible_Weight);
spNrQry.Boost = bst;
booleanquery.Add(spNrQry, Occur.SHOULD);
}
//This may need to be Should below
else if (i.boost == "Critical")
{
bst = (float)Convert.ToInt32(Critical_Weight);
//MessageBox.Show("The boost is: " + bst);
spNrQry.Boost = bst;
booleanquery.Add(spNrQry, Occur.SHOULD);
}

}
else //this must be a single term
{
//I am not sure if I should inclulde singletons here as it will give tf/idf unless I use the custom one where its reduced
}
}

//MessageBox.Show("The span terms are: " + booleanquery.ToString());

spResults = indexSearch.Search(booleanquery, spanMaxSize);//indexReader.MaxDoc);
int thsRes = spResults.TotalHits;
//MessageBox.Show("The number span terms are: " + thsRes.ToString());
// MessageBox.Show("This many spans: " + spResults.TotalHits.ToString());
return spResults;
}
catch (Exception err)
{
return spResults;
}
}