Example usage for edu.stanford.nlp.dcoref MentionExtractor MentionExtractor

Introduction

In this page you can find the example usage for edu.stanford.nlp.dcoref MentionExtractor MentionExtractor.

Prototype

public MentionExtractor(Dictionaries dict, Semantics semantics) 

Source Link

Usage

From source file:de.tudarmstadt.ukp.dkpro.core.stanfordnlp.StanfordCoreferenceResolver.java

License:Open Source License

@Override
public void initialize(UimaContext aContext) throws ResourceInitializationException {
    super.initialize(aContext);

    modelProvider = new ModelProviderBase<Coreferencer>() {
        {/*from ww w.  ja v  a2  s .  c o m*/
            setContextObject(StanfordCoreferenceResolver.this);

            setDefault(ARTIFACT_ID, "${groupId}.stanfordnlp-model-coref-${language}-${variant}");
            setDefault(LOCATION, "classpath:/${package}/lib/coref/${language}/${variant}/countries");
            setDefault(VARIANT, "default");

            // setOverride(LOCATION, modelLocation);
            // setOverride(LANGUAGE, language);
            // setOverride(VARIANT, variant);
        }

        @Override
        protected Coreferencer produceResource(URL aUrl) throws IOException {
            String base = FilenameUtils.getFullPathNoEndSeparator(aUrl.toString()) + "/";

            Properties props = new Properties();
            props.setProperty(Constants.SIEVES_PROP, sieves);
            props.setProperty(Constants.SCORE_PROP, String.valueOf(score));
            props.setProperty(Constants.POSTPROCESSING_PROP, String.valueOf(postprocessing));
            props.setProperty(Constants.SINGLETON_PROP, String.valueOf(singleton));
            props.setProperty(Constants.SINGLETON_MODEL_PROP, base + "singleton.predictor.ser");

            props.setProperty(Constants.MAXDIST_PROP, String.valueOf(maxdist));
            //              props.setProperty(Constants.BIG_GENDER_NUMBER_PROP, "false");
            props.setProperty(Constants.REPLICATECONLL_PROP, "false");
            props.setProperty(Constants.CONLL_SCORER, Constants.conllMentionEvalScript);

            // Cf. edu.stanford.nlp.dcoref.Dictionaries.Dictionaries(Properties)
            // props.getProperty(Constants.DEMONYM_PROP, DefaultPaths.DEFAULT_DCOREF_DEMONYM),
            props.setProperty(Constants.DEMONYM_PROP, base + "demonyms.txt");
            // props.getProperty(Constants.ANIMATE_PROP, DefaultPaths.DEFAULT_DCOREF_ANIMATE),
            props.setProperty(Constants.ANIMATE_PROP, base + "animate.unigrams.txt");
            // props.getProperty(Constants.INANIMATE_PROP, DefaultPaths.DEFAULT_DCOREF_INANIMATE),
            props.setProperty(Constants.INANIMATE_PROP, base + "inanimate.unigrams.txt");
            // props.getProperty(Constants.MALE_PROP),
            props.setProperty(Constants.MALE_PROP, base + "male.unigrams.txt");
            // props.getProperty(Constants.NEUTRAL_PROP),
            props.setProperty(Constants.NEUTRAL_PROP, base + "neutral.unigrams.txt");
            // props.getProperty(Constants.FEMALE_PROP),
            props.setProperty(Constants.FEMALE_PROP, base + "female.unigrams.txt");
            // props.getProperty(Constants.PLURAL_PROP),
            props.setProperty(Constants.PLURAL_PROP, base + "plural.unigrams.txt");
            // props.getProperty(Constants.SINGULAR_PROP),
            props.setProperty(Constants.SINGULAR_PROP, base + "singular.unigrams.txt");
            // props.getProperty(Constants.STATES_PROP, DefaultPaths.DEFAULT_DCOREF_STATES),
            props.setProperty(Constants.STATES_PROP, base + "state-abbreviations.txt");
            // props.getProperty(Constants.GENDER_NUMBER_PROP, DefaultPaths.DEFAULT_DCOREF_GENDER_NUMBER),
            props.setProperty(Constants.GENDER_NUMBER_PROP, base + "gender.map.ser.gz");
            // props.getProperty(Constants.COUNTRIES_PROP, DefaultPaths.DEFAULT_DCOREF_COUNTRIES),
            props.setProperty(Constants.COUNTRIES_PROP, base + "countries");
            // props.getProperty(Constants.STATES_PROVINCES_PROP, DefaultPaths.DEFAULT_DCOREF_STATES_AND_PROVINCES),
            props.setProperty(Constants.STATES_PROVINCES_PROP, base + "statesandprovinces");

            // The following properties are only relevant if the "CorefDictionaryMatch" sieve
            // is enabled.
            // PropertiesUtils.getStringArray(props, Constants.DICT_LIST_PROP,
            //   new String[]{DefaultPaths.DEFAULT_DCOREF_DICT1, DefaultPaths.DEFAULT_DCOREF_DICT2,
            //   DefaultPaths.DEFAULT_DCOREF_DICT3, DefaultPaths.DEFAULT_DCOREF_DICT4}),
            props.put(Constants.DICT_LIST_PROP, '[' + base + "coref.dict1.tsv" + ',' + base + "coref.dict1.tsv"
                    + ',' + base + "coref.dict1.tsv" + ',' + base + "coref.dict1.tsv" + ']');
            // props.getProperty(Constants.DICT_PMI_PROP, DefaultPaths.DEFAULT_DCOREF_DICT1),
            props.put(Constants.DICT_PMI_PROP, base + "coref.dict1.tsv");
            // props.getProperty(Constants.SIGNATURES_PROP, DefaultPaths.DEFAULT_DCOREF_NE_SIGNATURES));
            props.put(Constants.SIGNATURES_PROP, base + "ne.signatures.txt");

            try {
                Coreferencer coref = new Coreferencer();
                coref.corefSystem = new SieveCoreferenceSystem(props);
                coref.mentionExtractor = new MentionExtractor(coref.corefSystem.dictionaries(),
                        coref.corefSystem.semantics());
                return coref;
            } catch (Exception e) {
                throw new IOException(e);
            }
        }
    };
}

---