/// <summary>
/// Computes the similarity for the result object
/// </summary>
/// <param name="result">The result object, previously created with <see cref="Prepare(IEnumerable{HtmlNode}, string, List{string})" /></param>
/// <param name="output">The output.</param>
/// <param name="documents">Optional: select subset of documents to be analysed. These must be within <see cref="result" /> inner collections</param>
/// <returns>
/// The same result object specified in parameters
/// </returns>
public DocumentSimilarityResult ComputeSimilarity(DocumentSimilarityResult result, ITextRender output, List <HtmlNode> documents = null)
{
if (documents.isNullOrEmpty())
{
documents = result.LeafDictionaryByDocuments.Keys.ToList();
}
List <ComputeSimilarityTask> tasks = new List <ComputeSimilarityTask>();
for (int i = 0; i < documents.Count - 1; i++)
{
for (int y = i + 1; y < documents.Count; y++)
{
ComputeSimilarityTask task = new ComputeSimilarityTask()
{
documentA = documents[i],
documentB = documents[y],
nGrams_A = result.NGramsByDocuments[documents[i]],
nGrams_B = result.NGramsByDocuments[documents[y]]
};
tasks.Add(task);
//var documentA = ;
//var documentB = documents[y];
//var ABResult = ComputeSimilarity(documentA, documentB, result);
//result.AddResult(ABResult);
}
}
var task_chunks = tasks.SplitBySize((tasks.Count / 5));
foreach (var task_chunk in task_chunks)
{
output.AppendLine("Executing similarity computation task chunk [size:" + task_chunk.Count + "] " + task_chunks.IndexOf(task_chunk) + " of " + task_chunks.Count);
Parallel.ForEach <ComputeSimilarityTask>(task_chunk, x =>
{
ComputeSimilarity(x);
}
);
foreach (var task in task_chunk)
{
if (task.output != null)
{
result.AddResult(task.output);
}
}
}
return(result);
}
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;
}
