public static void Train(XMLToAnnotation.Data data, Properties props)
{
IDictionary <string, IList <Person> > characterMap = QuoteAttributionUtils.ReadPersonMap(props.GetProperty("charactersPath"));
IDictionary <int, string> pronounCorefMap = QuoteAttributionUtils.SetupCoref(props.GetProperty("booknlpCoref"), characterMap, data.doc);
ICollection <string> animacyList = QuoteAttributionUtils.ReadAnimacyList(QuoteAttributionAnnotator.AnimacyWordList);
SupervisedSieveTraining.FeaturesData fd = Featurize(new SupervisedSieveTraining.SieveData(data.doc, characterMap, pronounCorefMap, animacyList), data.goldList, true);
ExtractQuotesClassifier quotesClassifier = new ExtractQuotesClassifier(fd.dataset);
OutputModel(props.GetProperty("modelPath"), quotesClassifier.GetClassifier());
}
/// <exception cref="System.Exception"/>
public static void Main(string[] args)
{
string home = "/home/mjfang/action_grammars/";
// make the first argument one for a base directory
string specificFile = "1PPDevUncollapsed.props";
if (args.Length >= 1)
{
home = args[0];
}
if (args.Length >= 2)
{
specificFile = args[1];
}
System.Console.Out.WriteLine("Base directory: " + home);
Properties props = StringUtils.PropFileToProperties(home + "ExtractQuotesXMLScripts/" + specificFile);
XMLToAnnotation.Data data = XMLToAnnotation.ReadXMLFormat(props.GetProperty("file"));
Properties propsPara = new Properties();
propsPara.SetProperty("paragraphBreak", "one");
ParagraphAnnotator pa = new ParagraphAnnotator(propsPara, false);
pa.Annotate(data.doc);
Properties annotatorProps = new Properties();
annotatorProps.SetProperty("charactersPath", props.GetProperty("charactersPath"));
//"characterList.txt"
annotatorProps.SetProperty("booknlpCoref", props.GetProperty("booknlpCoref"));
annotatorProps.SetProperty("modelPath", props.GetProperty("modelPath"));
//"model.ser");
QuoteAttributionAnnotator qaa = new QuoteAttributionAnnotator(annotatorProps);
qaa.Annotate(data.doc);
ChapterAnnotator ca = new ChapterAnnotator();
ca.Annotate(data.doc);
Train(data, annotatorProps);
}
