public void Tag(Sentence sentence, TagOptions options)
{
// need training to generate model
if(_contextMapping == null)
{
var corpus = new CoNLLReader().Read(new ReaderOptions
{
DataDir = Path.Combine(options.CorpusDir, "CoNLL"),
FileName = "conll2000_chunking_train.txt"
});
Train(corpus, options);
}
Fill(sentence, options);
for (int pos = options.NGram - 1; pos < sentence.Words.Count; pos++)
{
sentence.Words[pos].Pos = _contextMapping.FirstOrDefault(x => x.Context == GetContext(pos, sentence.Words, options))?.Tag;
// set default tag
if(sentence.Words[pos].Pos == null)
{
sentence.Words[pos].Pos = options.Tag;
}
}
for(int pos = 0; pos < options.NGram - 1; pos++)
{
sentence.Words.RemoveAt(0);
}
}
public BotSharpTagger()
{
string dataDir = Path.Combine(AppDomain.CurrentDomain.GetData("DataPath").ToString(), "Corpus", "CoNLL");
var data = new CoNLLReader().Read(new ReaderOptions
{
DataDir = dataDir,
FileName = "conll2000_chunking_train.txt"
});
_tagger = new TaggerFactory <NGramTagger>(new TagOptions
{
NGram = 1,
Tag = "NN",
Corpus = data
}, SupportedLanguage.English);
}
