protected void ComputeSimilarity(ComputeSimilarityTask task)
{
var score_StructureSimilarity = StructureSimilarity.GetSimilarity(task.nGrams_A, task.nGrams_B, settings.computationMethod);
var score_ContentSimilarity = ContentSimilarity.GetSimilarity(task.nGrams_A, task.nGrams_B, settings.computationMethod);
DocumentSimilarityResultPair output = new DocumentSimilarityResultPair
{
itemA = task.documentA,
itemB = task.documentB,
StructureSimilarity = score_StructureSimilarity,
ContentSimilarity = score_ContentSimilarity
};
task.output = output;
}
/// <summary>
/// Computes similarity for two documents, that are part of result's inner collections
/// </summary>
/// <param name="documentA">The document a.</param>
/// <param name="documentB">The document b.</param>
/// <param name="result">The result object, previously created with <see cref="Prepare(IEnumerable{HtmlNode}, string, List{string})"/></param>
/// <returns>Result for these two documents</returns>
public DocumentSimilarityResultPair ComputeSimilarity(HtmlNode documentA, HtmlNode documentB, DocumentSimilarityResult result)
{
List <LeafNodeDictionaryEntryNGram> nGrams_A = result.NGramsByDocuments[documentA]; //setAnalysisTools<LeafNodeDictionaryEntry>.getNGramSet<LeafNodeDictionaryEntryNGram>(leafNodeDictionaryA.items, settings.nGramWidth, settings.nGramMode);
List <LeafNodeDictionaryEntryNGram> nGrams_B = result.NGramsByDocuments[documentB]; //setAnalysisTools<LeafNodeDictionaryEntry>.getNGramSet<LeafNodeDictionaryEntryNGram>(leafNodeDictionaryB.items, settings.nGramWidth, settings.nGramMode);
var score_StructureSimilarity = StructureSimilarity.GetSimilarity(nGrams_A, nGrams_B, settings.computationMethod);
var score_ContentSimilarity = ContentSimilarity.GetSimilarity(nGrams_A, nGrams_B, settings.computationMethod);
DocumentSimilarityResultPair output = new DocumentSimilarityResultPair
{
itemA = documentA,
itemB = documentB,
StructureSimilarity = score_StructureSimilarity,
ContentSimilarity = score_ContentSimilarity
};
return(output);
}
/// <summary>
/// Computes the similarity between two items. If you have to compare more than two documents, use <see cref="Prepare(IEnumerable{HtmlNode}, string, List{string})"/> and <see cref="ComputeSimilarity(HtmlNode, HtmlNode, DocumentSimilarityResult)"/> methods for greater performances
/// </summary>
/// <param name="documentA">The document a.</param>
/// <param name="documentB">The document b.</param>
/// <returns></returns>
public DocumentSimilarityResultPair ComputeSimilarity(HtmlNode documentA, HtmlNode documentB)
{
LeafNodeDictionary leafNodeDictionaryA = new LeafNodeDictionary(documentA);
LeafNodeDictionary leafNodeDictionaryB = new LeafNodeDictionary(documentB);
List <LeafNodeDictionaryEntryNGram> nGrams_A = setAnalysisTools <LeafNodeDictionaryEntry> .getNGramSet <LeafNodeDictionaryEntryNGram>(leafNodeDictionaryA.items, settings.nGramWidth, settings.nGramMode);
List <LeafNodeDictionaryEntryNGram> nGrams_B = setAnalysisTools <LeafNodeDictionaryEntry> .getNGramSet <LeafNodeDictionaryEntryNGram>(leafNodeDictionaryB.items, settings.nGramWidth, settings.nGramMode);
DocumentSimilarityResultPair output = new DocumentSimilarityResultPair
{
itemA = documentA,
itemB = documentB
};
output.ContentSimilarity = ContentSimilarity.GetSimilarity(nGrams_A, nGrams_B, settings.computationMethod);
output.StructureSimilarity = ContentSimilarity.GetSimilarity(nGrams_A, nGrams_B, settings.computationMethod);
return(output);
}
public void AddResult(DocumentSimilarityResultPair pair)
{
ContentSimilarityRange.Learn(pair.ContentSimilarity);
StructureSimilarityRange.Learn(pair.StructureSimilarity);
items.Add(pair);
}
