public MentionExtractor(Dictionaries dict, Semantics semantics)
{
this.headFinder = new SemanticHeadFinder();
this.dictionaries = dict;
this.semantics = semantics;
this.mentionFinder = new RuleBasedCorefMentionFinder();
}
/// <exception cref="System.Exception"/>
public MUCMentionExtractor(Dictionaries dict, Properties props, Semantics semantics)
: base(dict, semantics)
{
string fileName = props.GetProperty(Constants.MucProp);
fileContents = IOUtils.SlurpFile(fileName);
currentOffset = 0;
tokenizerFactory = PTBTokenizer.Factory(new CoreLabelTokenFactory(false), string.Empty);
stanfordProcessor = LoadStanfordProcessor(props);
}
/// <exception cref="System.Exception"/>
public CoNLLMentionExtractor(Dictionaries dict, Properties props, Semantics semantics)
: base(dict, semantics)
{
// Initialize reader for reading from CONLL2011 corpus
corpusPath = props.GetProperty(Constants.Conll2011Prop);
replicateCoNLL = bool.ParseBoolean(props.GetProperty(Constants.ReplicateconllProp, "false"));
CoNLL2011DocumentReader.Options options = new CoNLL2011DocumentReader.Options();
options.annotateTokenCoref = false;
options.annotateTokenSpeaker = Constants.UseGoldSpeakerTags || replicateCoNLL;
options.annotateTokenNer = Constants.UseGoldNe || replicateCoNLL;
options.annotateTokenPos = Constants.UseGoldPos || replicateCoNLL;
options.SetFilter(".*_auto_conll$");
reader = new CoNLL2011DocumentReader(corpusPath, options);
stanfordProcessor = LoadStanfordProcessor(props);
}
/// <exception cref="System.Exception"/>
public ACEMentionExtractor(Dictionaries dict, Properties props, Semantics semantics)
: base(dict, semantics)
{
stanfordProcessor = LoadStanfordProcessor(props);
if (props.Contains(Constants.Ace2004Prop))
{
corpusPath = props.GetProperty(Constants.Ace2004Prop);
aceReader = new AceReader(stanfordProcessor, false, "ACE2004");
}
else
{
if (props.Contains(Constants.Ace2005Prop))
{
corpusPath = props.GetProperty(Constants.Ace2005Prop);
aceReader = new AceReader(stanfordProcessor, false);
}
}
aceReader.SetLoggerLevel(Level.Info);
if (corpusPath[corpusPath.Length - 1] != File.separatorChar)
{
corpusPath += File.separatorChar;
}
files = new File(corpusPath).List();
}
/// <exception cref="System.Exception"/>
public CoNLLMentionExtractor(Dictionaries dict, Properties props, Semantics semantics, LogisticClassifier <string, string> singletonModel)
: this(dict, props, semantics)
{
singletonPredictor = singletonModel;
}
