public override ObjectStream <NameSample> create(string[] args)
{
Parameters @params = ArgumentParser.parse(args, typeof(Parameters));
TokenizerModel tokenizerModel = (new TokenizerModelLoader()).load(@params.TokenizerModel);
Tokenizer tokenizer = new TokenizerME(tokenizerModel);
ObjectStream <string> mucDocStream = new FileToStringSampleStream(new DirectorySampleStream(@params.Data, new FileFilterAnonymousInnerClassHelper(this), false), Charset.forName("UTF-8"));
return(new MucNameSampleStream(tokenizer, mucDocStream));
}
public override ObjectStream <CorefSample> create(string[] args)
{
Parameters @params = ArgumentParser.parse(args, typeof(Parameters));
ParserModel parserModel = (new ParserModelLoader()).load(@params.ParserModel);
Parser parser = ParserFactory.create(parserModel);
TokenizerModel tokenizerModel = (new TokenizerModelLoader()).load(@params.TokenizerModel);
Tokenizer tokenizer = new TokenizerME(tokenizerModel);
ObjectStream <string> mucDocStream = new FileToStringSampleStream(new DirectorySampleStream(@params.Data, new FileFilterAnonymousInnerClassHelper(this), false), Charset.forName("UTF-8"));
ObjectStream <RawCorefSample> rawSamples = new MucCorefSampleStream(tokenizer, mucDocStream);
ObjectStream <RawCorefSample> parsedSamples = new FullParseCorefEnhancerStream(parser, rawSamples);
// How to load all these nameFinder models ?!
// Lets make a param per model, not that nice, but ok!
IDictionary <string, Jfile> modelFileTagMap = new Dictionary <string, Jfile>();
modelFileTagMap["person"] = @params.PersonModel;
modelFileTagMap["organization"] = @params.OrganizationModel;
IList <TokenNameFinder> nameFinders = new List <TokenNameFinder>();
IList <string> tags = new List <string>();
foreach (KeyValuePair <string, Jfile> entry in modelFileTagMap)
{
nameFinders.Add(new NameFinderME((new TokenNameFinderModelLoader()).load(entry.Value)));
tags.Add(entry.Key);
}
return(new MucMentionInserterStream(new NameFinderCorefEnhancerStream(nameFinders.ToArray(), tags.ToArray(), parsedSamples)));
}
