public double TFIDFSimilarity(Document <T> query, Corpus <T> corpus)
{
double[] queryWeights = corpus.Select(term => query.TFIDF(term, corpus)).ToArray();
double[] documentWeights = corpus.Select(term => TFIDF(term, corpus)).ToArray();
var vectorSpaceModel = new VectorSpaceModel(queryWeights, documentWeights);
return(vectorSpaceModel.CalculateCosineSimilarity());
}
public double TFIDF(T term, Corpus <T> corpus)
{
return(TFIDF(term, corpus, AugmentedTermFrequency));
}
public double TFIDF(T term, Corpus <T> corpus, Func <T, double> termFrequencyFunction)
{
return(termFrequencyFunction(term) * corpus.IDF(term));
}