private Dictionary <int, double> CalculateSentenceScores(Dictionary <string, double> normalizedTermFreqInverseSentFreqValues)
{
Dictionary <int, double> rankedSentence = new Dictionary <int, double>();
for (int i = 0; i < _text.GetSentenceCount(); i++)
{
double sentenceScore = 0;
for (int k = 0; k < _text.GetWordCountInSentence(i); k++)
{
sentenceScore += normalizedTermFreqInverseSentFreqValues[_text.GetWord(i, k)];
}
rankedSentence[i] = sentenceScore;
}
//normalize the sentence scores
var highestSentenceScore = rankedSentence.Values.Max();
for (int i = 0; i < _text.GetSentenceCount(); i++)
{
rankedSentence[i] = rankedSentence[i] / highestSentenceScore;
}
return(rankedSentence);
}
/// <summary>
/// Handles the Cue Phrase Feature
/// </summary>
/// <param name="cuePhraseList"></param>
/// <param name="scoreWeighting"></param>
/// <returns></returns>
private Dictionary <int, double> RunCuePhraseFeature(List <string> cuePhraseList, double scoreWeighting)
{
Dictionary <int, double> sentenceScore = new Dictionary <int, double>();
for (int sentenceNumber = 0; sentenceNumber < _unstemmedText.GetSentenceCount(); sentenceNumber++)
{
var sentence = _unstemmedText.GetSentence(sentenceNumber);
string formedSentence = string.Join(" ", sentence);
var cuePhrasesInSentence = cuePhraseList.Any(phrase => formedSentence.Contains(phrase));
//if there is a cue phrase exists, add the weighting score to the sentence (per each word)
if (cuePhrasesInSentence)
{
//todo: tweakable point to improve algortihm
// the score is added to each word in the sentence per every cue phrase in the sentence.
double scoreToAdd = scoreWeighting * _unstemmedText.GetWordCountInSentence(sentenceNumber);
sentenceScore[sentenceNumber] = scoreToAdd;
}
else
{
sentenceScore[sentenceNumber] = 0;
}
}
return(sentenceScore);
}
public static TextModel WordStemmer(TextModel text)
{
Stemmer stemmer = new Stemmer();
for (int i = 0; i < text.GetSentenceCount(); i++)
{
for (int k = 0; k < text.GetWordCountInSentence(i); k++)
{
var stemmedWord = stemmer.StemWord(text.GetWord(i, k));
text.SetWord(i, k, stemmedWord);
}
}
return(text);
}