public void TestContains()
{
Assert.True(corpus.Contains("atatürk"));
foreach (Word word in corpus.GetWordList())
{
Assert.True(corpus.Contains(word.GetName()));
}
Assert.True(simpleCorpus.Contains("mehmet"));
foreach (Word word in simpleCorpus.GetWordList())
{
Assert.True(simpleCorpus.Contains(word.GetName()));
}
}
/**
* <summary>Constructor for the {@link Vocabulary} class. For each distinct word in the corpus, a {@link VocabularyWord}
* instance is created. After that, words are sorted according to their occurrences. Unigram table is constructed,
* where after Huffman tree is created based on the number of occurrences of the words.</summary>
* <param name="corpus">Corpus used to train word vectors using Word2Vec algorithm.</param>
*/
public Vocabulary(Corpus.Corpus corpus)
{
var wordList = corpus.GetWordList();
_vocabulary = new List <VocabularyWord>();
foreach (var word in wordList)
{
_vocabulary.Add(new VocabularyWord(word.GetName(), corpus.GetCount(word)));
}
_vocabulary.Sort();
CreateUniGramTable();
ConstructHuffmanTree();
_vocabulary.Sort(new TurkishWordComparator());
}
