public void BuildVocab(string corpus)
{
Debug.WriteLine($"{DateTime.Now:yyyy-MM-dd hh:mm:ss.fffff} {nameof(Word2Vec)} Start BuildVocab");
AssignCorpus(corpus);
var vocab = _allText.Distinct().ToList();
var dict = new Dictionary <string, int>();
foreach (var v in vocab)
{
if (String.IsNullOrWhiteSpace(v))
{
continue;
}
var cn = _allText.Count(a => a == v);
dict.Add(v, cn);
}
Vocab = new HuffmanEncoding(dict);
Debug.WriteLine($"{DateTime.Now:yyyy-MM-dd hh:mm:ss.fffff} {nameof(Word2Vec)} End BuildVocab");
}
public void BuildVocab(Dictionary <string, int> vocab)
{
Vocab = new HuffmanEncoding(vocab);
}