public PosTaggerProcessorConfig(TokenizerConfig tokenizerConfig,
string posTaggerResourcesXmlFilename)
{
Model = new PosTaggerResourcesModel(posTaggerResourcesXmlFilename);
TokenizerConfig = tokenizerConfig;
//set pos-tagger specially
TokenizerConfig.TokenizeMode |= TokenizeMode.PosTagger;
if (TokenizerConfig.PosTaggerInputTypeProcessorFactory == null)
{
TokenizerConfig.PosTaggerInputTypeProcessorFactory = new PosTaggerInputTypeProcessorFactory(Model, TokenizerConfig.LanguageType);
}
}
public PosTaggerProcessorConfig(string tokenizerResourcesXmlFilename,
string posTaggerResourcesXmlFilename,
LanguageTypeEnum languageType,
SentSplitterConfig sentSplitterConfig)
{
Model = new PosTaggerResourcesModel(posTaggerResourcesXmlFilename);
TokenizerConfig = new TokenizerConfig(tokenizerResourcesXmlFilename)
{
TokenizeMode = TokenizeMode.PosTagger,
LanguageType = languageType,
SentSplitterConfig = sentSplitterConfig,
PosTaggerInputTypeProcessorFactory = new PosTaggerInputTypeProcessorFactory(Model, languageType),
};
}
internal PosTaggerInputTypeProcessorFactory(PosTaggerResourcesModel model, LanguageTypeEnum languageType)
{
switch (languageType)
{
case LanguageTypeEnum.Ru:
_PosTaggerInputTypeProcessor = new PosTaggerInputTypeProcessor_Ru(model.Numbers, model.Abbreviations);
break;
case LanguageTypeEnum.En:
_PosTaggerInputTypeProcessor = new PosTaggerInputTypeProcessor_En(model.Numbers, model.Abbreviations);
break;
default:
throw (new ArgumentException(languageType.ToString()));
}
}
public PosTaggerPreMerging(PosTaggerResourcesModel model)
{
_Model = model;
}
