public void AddDocument(Document document)
{
foreach (string word in document.Words)
{
_words.Add(word);
}
}
public Document PureDocument(Document document)
{
var words = new List<string>();
foreach (var word in document.Words)
{
if(_termsCollection.ContainsTerm(word))
words.Add(word);
}
return new Document(words);
}
/// <summary>
/// Stems document, creating new document, which is stemmed
/// </summary>
/// <param name="document">document to stem</param>
/// <param name="stemmerInterface">stemmer interface to use</param>
/// <returns>document, with all words stemmed.</returns>
public static Document StemDocument(Document document, IStemmerInterface stemmerInterface)
{
if (document == null) throw new ArgumentNullException("document");
var words = new List<string>();
foreach (string word in document.Words)
{
words.Add(stemmerInterface.StemTerm(word));
}
return new Document(words);
}
public BagOfWords(Document document)
{
foreach (string word in document.Words)
{
if (_wordCount.ContainsKey(word))
{
_wordCount[word]++;
}
else
{
_wordCount[word] = 1;
}
}
}
public IEnumerable<SearchResult> Search(Document query)
{
var queryTfIdf = new TfIdf(new Tf(new BagOfWords(query)), _idf);
var results = new List<SearchResult>();
foreach (TfIdfWithDocument tfIdfWithDocument in _tfIdfWithDocuments)
{
double probability = _probabilityMatrixCalculator.CalculateProbability(queryTfIdf,
tfIdfWithDocument.TfIdf,
_termsCollection);
results.Add(new SearchResult(tfIdfWithDocument.Document, probability));
}
return results.OrderByDescending(x => x.Probability);
}
public TfIdfWithDocument(TfIdf tfIdf, Document document)
{
TfIdf = tfIdf;
Document = document;
}
