Examples with Stanford331Configurator edu.illinois.cs.cogcomp.pipeline.common.Stanford331Configurator used on opensource projects

Example 1 with Stanford331Configurator

use of edu.illinois.cs.cogcomp.pipeline.common.Stanford331Configurator in project cogcomp-nlp by CogComp.

the class PipelineFactory method buildPipeline.

/**
     * create an AnnotatorService with components specified by the ResourceManager (to override
     * defaults in {@link PipelineConfigurator}
     * 
     * @param rm non-default config options
     * @return AnnotatorService with specified NLP components
     * @throws IOException
     * @throws AnnotatorException
     */
public static BasicAnnotatorService buildPipeline(ResourceManager rm) throws IOException, AnnotatorException {
    // Merges default configuration with the user-specified overrides.
    ResourceManager fullRm = (new PipelineConfigurator()).getConfig(new Stanford331Configurator().getConfig(rm));
    Boolean splitOnDash = fullRm.getBoolean(PipelineConfigurator.SPLIT_ON_DASH);
    boolean isSentencePipeline = fullRm.getBoolean(PipelineConfigurator.USE_SENTENCE_PIPELINE.key);
    if (isSentencePipeline) {
        // update cache directory to be distinct from regular pipeline
        String cacheDir = fullRm.getString(AnnotatorServiceConfigurator.CACHE_DIR.key);
        cacheDir += "_sentence";
        Properties props = fullRm.getProperties();
        props.setProperty(AnnotatorServiceConfigurator.CACHE_DIR.key, cacheDir);
        fullRm = new ResourceManager(props);
    }
    TextAnnotationBuilder taBldr = new TokenizerTextAnnotationBuilder(new StatefulTokenizer(splitOnDash));
    Map<String, Annotator> annotators = buildAnnotators(fullRm);
    return isSentencePipeline ? new SentencePipeline(taBldr, annotators, fullRm) : new BasicAnnotatorService(taBldr, annotators, fullRm);
}

Example 2 with Stanford331Configurator

use of edu.illinois.cs.cogcomp.pipeline.common.Stanford331Configurator in project cogcomp-nlp by CogComp.

the class PipelineFactory method buildPipeline.

/**
     * create an AnnotatorService with the given view names in the argument. The names are supposed
     * be strings, separated by space.
     *
     * @return AnnotatorService with specified NLP components
     * @throws IOException
     * @throws AnnotatorException
     */
public static BasicAnnotatorService buildPipeline(Boolean disableCache, String... views) throws IOException, AnnotatorException {
    List<String> allViewNames = ViewNames.getAllViewNames();
    Map<String, String> nonDefaultValues = new HashMap<>();
    for (String vu : views) {
        if (allViewNames.contains(vu)) {
            switch(vu) {
                case ViewNames.POS:
                    nonDefaultValues.put(PipelineConfigurator.USE_POS.key, Configurator.TRUE);
                    break;
                case ViewNames.LEMMA:
                    nonDefaultValues.put(PipelineConfigurator.USE_LEMMA.key, Configurator.TRUE);
                    break;
                case ViewNames.NER_CONLL:
                    nonDefaultValues.put(PipelineConfigurator.USE_NER_CONLL.key, Configurator.TRUE);
                    break;
                case ViewNames.NER_ONTONOTES:
                    nonDefaultValues.put(PipelineConfigurator.USE_NER_ONTONOTES.key, Configurator.TRUE);
                    break;
                case ViewNames.QUANTITIES:
                    nonDefaultValues.put(PipelineConfigurator.USE_QUANTIFIER.key, Configurator.TRUE);
                    break;
                case ViewNames.SHALLOW_PARSE:
                    nonDefaultValues.put(PipelineConfigurator.USE_SHALLOW_PARSE.key, Configurator.TRUE);
                    break;
                case ViewNames.SRL_VERB:
                    nonDefaultValues.put(PipelineConfigurator.USE_SRL_VERB.key, Configurator.TRUE);
                    break;
                case ViewNames.DEPENDENCY_STANFORD:
                    nonDefaultValues.put(PipelineConfigurator.USE_STANFORD_DEP.key, Configurator.TRUE);
                    break;
                case ViewNames.DEPENDENCY:
                    nonDefaultValues.put(PipelineConfigurator.USE_DEP.key, Configurator.TRUE);
                    break;
                case ViewNames.PARSE_STANFORD:
                    nonDefaultValues.put(PipelineConfigurator.USE_STANFORD_PARSE.key, Configurator.TRUE);
                    break;
                case ViewNames.SRL_PREP:
                    nonDefaultValues.put(PipelineConfigurator.USE_SRL_PREP.key, Configurator.TRUE);
                    break;
                case ViewNames.SRL_COMMA:
                    nonDefaultValues.put(PipelineConfigurator.USE_SRL_COMMA.key, Configurator.TRUE);
                    break;
                default:
                    logger.warn("View name " + vu + " is not supported yet. Look into the readme of the pipeline to see the list of valid annotators. ");
            }
        } else {
            throw new IllegalArgumentException("The view name " + vu + " is not a valid view name. " + "The possible view names are static members of the class `ViewName`. ");
        }
    }
    if (disableCache) {
        nonDefaultValues.put(AnnotatorServiceConfigurator.DISABLE_CACHE.key, Configurator.TRUE);
    } else {
        nonDefaultValues.put(AnnotatorServiceConfigurator.DISABLE_CACHE.key, Configurator.FALSE);
    }
    // using the default settings and changing the views
    ResourceManager fullRm = (new PipelineConfigurator()).getConfig(new Stanford331Configurator().getConfig(nonDefaultValues));
    boolean splitOnHypen = fullRm.getBoolean(PipelineConfigurator.SPLIT_ON_DASH.key);
    TextAnnotationBuilder taBldr = new TokenizerTextAnnotationBuilder(new StatefulTokenizer(splitOnHypen));
    Map<String, Annotator> annotators = buildAnnotators(fullRm);
    return new SentencePipeline(taBldr, annotators, fullRm);
}

Example 3 with Stanford331Configurator

use of edu.illinois.cs.cogcomp.pipeline.common.Stanford331Configurator in project cogcomp-nlp by CogComp.

the class PipelineFactory method buildAnnotators.

/**
     * instantiate a set of annotators for use in an AnnotatorService object by default, will use
     * lazy initialization where possible -- change this behavior with the
     * {@link PipelineConfigurator#USE_LAZY_INITIALIZATION} property.
     * 
     * @param nonDefaultRm ResourceManager with all non-default values for Annotators
     * @return a Map from annotator view name to annotator
     */
private static Map<String, Annotator> buildAnnotators(ResourceManager nonDefaultRm) throws IOException {
    ResourceManager rm = new PipelineConfigurator().getConfig(new Stanford331Configurator().getConfig(nonDefaultRm));
    String timePerSentence = rm.getString(Stanford331Configurator.STFRD_TIME_PER_SENTENCE);
    String maxParseSentenceLength = rm.getString(Stanford331Configurator.STFRD_MAX_SENTENCE_LENGTH);
    boolean useLazyInitialization = rm.getBoolean(PipelineConfigurator.USE_LAZY_INITIALIZATION.key, PipelineConfigurator.TRUE);
    Map<String, Annotator> viewGenerators = new HashMap<>();
    if (rm.getBoolean(PipelineConfigurator.USE_POS)) {
        POSAnnotator pos = new POSAnnotator();
        viewGenerators.put(pos.getViewName(), pos);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_LEMMA)) {
        IllinoisLemmatizer lem = new IllinoisLemmatizer(rm);
        viewGenerators.put(lem.getViewName(), lem);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_SHALLOW_PARSE)) {
        viewGenerators.put(ViewNames.SHALLOW_PARSE, new ChunkerAnnotator());
    }
    if (rm.getBoolean(PipelineConfigurator.USE_NER_CONLL)) {
        NERAnnotator nerConll = NerAnnotatorManager.buildNerAnnotator(rm, ViewNames.NER_CONLL);
        viewGenerators.put(nerConll.getViewName(), nerConll);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_NER_ONTONOTES)) {
        NERAnnotator nerOntonotes = NerAnnotatorManager.buildNerAnnotator(rm, ViewNames.NER_ONTONOTES);
        viewGenerators.put(nerOntonotes.getViewName(), nerOntonotes);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_DEP)) {
        DepAnnotator dep = new DepAnnotator();
        viewGenerators.put(dep.getViewName(), dep);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_STANFORD_DEP) || rm.getBoolean(PipelineConfigurator.USE_STANFORD_PARSE)) {
        Properties stanfordProps = new Properties();
        stanfordProps.put("annotators", "pos, parse");
        stanfordProps.put("parse.originalDependencies", true);
        stanfordProps.put("parse.maxlen", maxParseSentenceLength);
        // per sentence? could be per
        stanfordProps.put("parse.maxtime", timePerSentence);
        // document but no idea from
        // stanford javadoc
        POSTaggerAnnotator posAnnotator = new POSTaggerAnnotator("pos", stanfordProps);
        ParserAnnotator parseAnnotator = new ParserAnnotator("parse", stanfordProps);
        int maxLength = Integer.parseInt(maxParseSentenceLength);
        boolean throwExceptionOnSentenceLengthCheck = rm.getBoolean(Stanford331Configurator.THROW_EXCEPTION_ON_FAILED_LENGTH_CHECK.key);
        if (rm.getBoolean(PipelineConfigurator.USE_STANFORD_DEP)) {
            StanfordDepHandler depParser = new StanfordDepHandler(posAnnotator, parseAnnotator, maxLength, throwExceptionOnSentenceLengthCheck);
            viewGenerators.put(depParser.getViewName(), depParser);
        }
        if (rm.getBoolean(PipelineConfigurator.USE_STANFORD_PARSE)) {
            StanfordParseHandler parser = new StanfordParseHandler(posAnnotator, parseAnnotator, maxLength, throwExceptionOnSentenceLengthCheck);
            viewGenerators.put(parser.getViewName(), parser);
        }
    }
    if (rm.getBoolean(PipelineConfigurator.USE_SRL_VERB)) {
        Properties verbProps = new Properties();
        String verbType = SRLType.Verb.name();
        verbProps.setProperty(SrlConfigurator.SRL_TYPE.key, verbType);
        ResourceManager verbRm = new ResourceManager(verbProps);
        rm = Configurator.mergeProperties(rm, verbRm);
        try {
            SemanticRoleLabeler verbSrl = new SemanticRoleLabeler(rm, useLazyInitialization);
            viewGenerators.put(ViewNames.SRL_VERB, verbSrl);
        } catch (Exception e) {
            throw new IOException("SRL verb cannot init: " + e.getMessage());
        }
    }
    if (rm.getBoolean(PipelineConfigurator.USE_SRL_NOM)) {
        Properties nomProps = new Properties();
        String nomType = SRLType.Nom.name();
        nomProps.setProperty(SrlConfigurator.SRL_TYPE.key, nomType);
        ResourceManager nomRm = new ResourceManager(nomProps);
        rm = Configurator.mergeProperties(rm, nomRm);
        try {
            SemanticRoleLabeler nomSrl = new SemanticRoleLabeler(rm, useLazyInitialization);
            // note that you can't call nomSrl (or verbSrl).getViewName() as it may not be
            // initialized yet
            viewGenerators.put(ViewNames.SRL_NOM, nomSrl);
        // viewGenerators.put(ViewNames.SRL_NOM,new SrlHandler("NomSRL", "5.1.9", nomType,
        // ViewNames.SRL_NOM,
        // useLazyInitialization, rm));
        } catch (Exception e) {
            throw new IOException("SRL nom cannot init .." + e.getMessage());
        }
    }
    if (rm.getBoolean(PipelineConfigurator.USE_QUANTIFIER)) {
        Quantifier quantifierAnnotator = new Quantifier();
        viewGenerators.put(ViewNames.QUANTITIES, quantifierAnnotator);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_SRL_PREP)) {
        PrepSRLAnnotator prepSRLAnnotator = new PrepSRLAnnotator();
        viewGenerators.put(ViewNames.SRL_PREP, prepSRLAnnotator);
    }
    if (rm.getBoolean(PipelineConfigurator.USE_SRL_COMMA)) {
        CommaLabeler commaLabeler = new CommaLabeler();
        viewGenerators.put(ViewNames.SRL_COMMA, commaLabeler);
    }
    return viewGenerators;
}