public List<SearchResultEntry> DoSearch(string searchTxt)
{
SearchText = searchTxt;
WordSet searchWordSet = new WordSet(searchTxt);
List<SearchResultEntry> answ = new List<SearchResultEntry>();
WordSetPerTag = new Dictionary<object, WordSet>();
WordSet currTextWs;
SearchResultEntry se;
foreach (TaggedText ttxt in TextsToBeSearched)
{
currTextWs = new WordSet(ttxt.Text);
WordSetPerTag.Add(ttxt.Tag, currTextWs);
se = CalculateGrade(currTextWs, searchWordSet);
se.TagMark = ttxt.Tag;
if(se.MatchGrade>=MinimumMatchGrade)
answ.Add(se);
}
answ.Sort();
return answ;
}
/// <summary>
/// Calculate the grade of equality for two word sets
/// </summary>
/// <param name="currTextWs"></param>
/// <param name="searchWordSet"></param>
/// <returns>A search result entry containing information about the found elements</returns>
private SearchResultEntry CalculateGrade(WordSet ws1, WordSet ws2)
{
SearchResultEntry answ = new SearchResultEntry();
answ.MatchTexts = new List<string>();
answ.CounterMatchTexts = new List<string>();
answ.MatchGrades = new List<double>();
answ.MatchPositions = new List<int>();
double gradeSum = 0.0;
int minldist, currldist, minidx;
int matchIdx;
string w1, minCounterMatch;
double currGrade;
foreach (string w2 in ws2.Words)
{
matchIdx = ws1.Words.IndexOf(w2);
if (matchIdx>=0)
{
answ.MatchTexts.Add(w2);
answ.CounterMatchTexts.Add(w2);
answ.MatchGrades.Add(1.0);
answ.MatchPositions.Add(ws1.Positions[matchIdx]);
gradeSum += 1.0;
}
else
{
minldist = int.MaxValue;
minCounterMatch = null;
minidx = 0;
for (int i=0; i<ws1.Words.Count; i++)
{
w1 = ws1.Words[i];
currldist = LevenshteinDistance(w1, w2);
if (currldist < minldist)
{
minldist = currldist;
minidx = i;
minCounterMatch = w1;
}
}
currGrade = (double)(w2.Length - minldist) / w2.Length;
if (currGrade >= MinimumMatchGrade)
{
answ.MatchTexts.Add(w2);
answ.MatchGrades.Add(currGrade);
answ.CounterMatchTexts.Add(minCounterMatch);
answ.MatchPositions.Add(ws1.Positions[minidx]);
gradeSum += currGrade;
}
}
}
answ.MatchGrade = gradeSum / (double)ws2.Words.Count;
return answ;
}