Examples with Ontology info.ephyra.nlp.semantics.ontologies.Ontology used on opensource projects

Example 1 with Ontology

use of info.ephyra.nlp.semantics.ontologies.Ontology in project lucida by claritylab.

the class OpenEphyra method initFactoid.

/**
	 * Initializes the pipeline for factoid questions.
	 */
protected void initFactoid() {
    // question analysis
    Ontology wordNet = new WordNet();
    // - dictionaries for term extraction
    QuestionAnalysis.clearDictionaries();
    QuestionAnalysis.addDictionary(wordNet);
    // - ontologies for term expansion
    QuestionAnalysis.clearOntologies();
    QuestionAnalysis.addOntology(wordNet);
    // query generation
    QueryGeneration.clearQueryGenerators();
    QueryGeneration.addQueryGenerator(new BagOfWordsG());
    QueryGeneration.addQueryGenerator(new BagOfTermsG());
    QueryGeneration.addQueryGenerator(new PredicateG());
    QueryGeneration.addQueryGenerator(new QuestionInterpretationG());
    QueryGeneration.addQueryGenerator(new QuestionReformulationG());
    // search
    // - knowledge miners for unstructured knowledge sources
    Search.clearKnowledgeMiners();
    //		Search.addKnowledgeMiner(new YahooKM());
    for (String[] indriIndices : IndriKM.getIndriIndices()) Search.addKnowledgeMiner(new IndriKM(indriIndices, false));
    //		for (String[] indriServers : IndriKM.getIndriServers())
    //			Search.addKnowledgeMiner(new IndriKM(indriServers, true));
    // - knowledge annotators for (semi-)structured knowledge sources
    Search.clearKnowledgeAnnotators();
    // answer extraction and selection
    // (the filters are applied in this order)
    AnswerSelection.clearFilters();
    // - answer extraction filters
    AnswerSelection.addFilter(new AnswerTypeFilter());
    AnswerSelection.addFilter(new AnswerPatternFilter());
    //AnswerSelection.addFilter(new WebDocumentFetcherFilter());
    AnswerSelection.addFilter(new PredicateExtractionFilter());
    AnswerSelection.addFilter(new FactoidsFromPredicatesFilter());
    AnswerSelection.addFilter(new TruncationFilter());
    // - answer selection filters
    AnswerSelection.addFilter(new StopwordFilter());
    AnswerSelection.addFilter(new QuestionKeywordsFilter());
    AnswerSelection.addFilter(new ScoreNormalizationFilter(NORMALIZER));
    AnswerSelection.addFilter(new ScoreCombinationFilter());
    AnswerSelection.addFilter(new FactoidSubsetFilter());
    AnswerSelection.addFilter(new DuplicateFilter());
    AnswerSelection.addFilter(new ScoreSorterFilter());
}

Example 2 with Ontology

use of info.ephyra.nlp.semantics.ontologies.Ontology in project lucida by claritylab.

the class OpenEphyraCorpus method initFactoidCorpus.

/**
	 * Initializes the pipeline for factoid questions, using a local corpus as a
	 * knowledge source.
	 */
protected void initFactoidCorpus() {
    // question analysis
    Ontology wordNet = new WordNet();
    // - dictionaries for term extraction
    QuestionAnalysis.clearDictionaries();
    QuestionAnalysis.addDictionary(wordNet);
    // - ontologies for term expansion
    QuestionAnalysis.clearOntologies();
    QuestionAnalysis.addOntology(wordNet);
    // query generation
    QueryGeneration.clearQueryGenerators();
    QueryGeneration.addQueryGenerator(new BagOfWordsG());
    QueryGeneration.addQueryGenerator(new BagOfTermsG());
    QueryGeneration.addQueryGenerator(new PredicateG());
    QueryGeneration.addQueryGenerator(new QuestionInterpretationG());
    QueryGeneration.addQueryGenerator(new QuestionReformulationG());
    // search
    // - knowledge miners for unstructured knowledge sources
    Search.clearKnowledgeMiners();
    for (String[] indriIndices : IndriKM.getIndriIndices()) Search.addKnowledgeMiner(new IndriKM(indriIndices, false));
    for (String[] indriServers : IndriKM.getIndriServers()) Search.addKnowledgeMiner(new IndriKM(indriServers, true));
    // - knowledge annotators for (semi-)structured knowledge sources
    Search.clearKnowledgeAnnotators();
    // answer extraction and selection
    // (the filters are applied in this order)
    AnswerSelection.clearFilters();
    // - answer extraction filters
    AnswerSelection.addFilter(new AnswerTypeFilter());
    AnswerSelection.addFilter(new AnswerPatternFilter());
    AnswerSelection.addFilter(new WebDocumentFetcherFilter());
    AnswerSelection.addFilter(new PredicateExtractionFilter());
    AnswerSelection.addFilter(new FactoidsFromPredicatesFilter());
    AnswerSelection.addFilter(new TruncationFilter());
// - answer selection filters
}

Example 3 with Ontology

use of info.ephyra.nlp.semantics.ontologies.Ontology in project lucida by claritylab.

the class OpenEphyraCorpus method initFactoidWeb.

/**
	 * Initializes the pipeline for factoid questions, using the Web as a
	 * knowledge source.
	 * 
	 * @param resultsCorp results retrieved from the corpus
	 */
protected void initFactoidWeb(Result[] resultsCorp) {
    // question analysis
    Ontology wordNet = new WordNet();
    // - dictionaries for term extraction
    QuestionAnalysis.clearDictionaries();
    QuestionAnalysis.addDictionary(wordNet);
    // - ontologies for term expansion
    QuestionAnalysis.clearOntologies();
    QuestionAnalysis.addOntology(wordNet);
    // query generation
    QueryGeneration.clearQueryGenerators();
    QueryGeneration.addQueryGenerator(new BagOfWordsG());
    QueryGeneration.addQueryGenerator(new BagOfTermsG());
    QueryGeneration.addQueryGenerator(new PredicateG());
    QueryGeneration.addQueryGenerator(new QuestionInterpretationG());
    QueryGeneration.addQueryGenerator(new QuestionReformulationG());
    // search
    // - knowledge miners for unstructured knowledge sources
    Search.clearKnowledgeMiners();
    Search.addKnowledgeMiner(new BingKM());
    //		Search.addKnowledgeMiner(new GoogleKM());
    //		Search.addKnowledgeMiner(new YahooKM());
    // - knowledge annotators for (semi-)structured knowledge sources
    Search.clearKnowledgeAnnotators();
    // answer extraction and selection
    // (the filters are applied in this order)
    AnswerSelection.clearFilters();
    // - answer extraction filters
    AnswerSelection.addFilter(new AnswerTypeFilter());
    AnswerSelection.addFilter(new AnswerPatternFilter());
    AnswerSelection.addFilter(new WebDocumentFetcherFilter());
    AnswerSelection.addFilter(new PredicateExtractionFilter());
    AnswerSelection.addFilter(new FactoidsFromPredicatesFilter());
    AnswerSelection.addFilter(new TruncationFilter());
    // - answer selection filters
    AnswerSelection.addFilter(new StopwordFilter());
    AnswerSelection.addFilter(new QuestionKeywordsFilter());
    AnswerSelection.addFilter(new AnswerProjectionFilter(resultsCorp));
    AnswerSelection.addFilter(new ScoreNormalizationFilter(NORMALIZER));
    AnswerSelection.addFilter(new ScoreCombinationFilter());
    AnswerSelection.addFilter(new FactoidSubsetFilter());
    AnswerSelection.addFilter(new DuplicateFilter());
    AnswerSelection.addFilter(new ScoreSorterFilter());
    AnswerSelection.addFilter(new ResultLengthFilter());
}

Example 4 with Ontology

use of info.ephyra.nlp.semantics.ontologies.Ontology in project lucida by claritylab.

the class TermExpander method expandTerm.

/**
	 * Expands a term by looking up related terms in ontologies.
	 * 
	 * @param term a term
	 * @param ps predicates in the same sentence
	 * @param ontologies ontologies used to expand the term
	 */
public static void expandTerm(Term term, Predicate[] ps, Ontology[] ontologies) {
    String text = term.getText();
    String pos = term.getPos();
    Map<String, Double> lemmas = new Hashtable<String, Double>();
    Map<String, Double> expansions = new Hashtable<String, Double>();
    // expand events, entities and modifiers
    if (isTarget(term, ps) || pos.startsWith("VB")) {
        // lemmatize verbs that are in WordNet
        String lemma = WordNet.getLemma(text, POS.VERB);
        if (lemma == null)
            lemma = text;
        // set lemma if the POS was misleading
        if (!pos.startsWith("VB"))
            term.setLemma(lemma);
        // expand event
        for (Ontology ontology : ontologies) {
            Map<String, Double> expanded = ontology.expandEvent(lemma);
            lemmas.putAll(expanded);
        }
        // ensure that there are at most MAX_EXPANSIONS expansions with
        // weights of at least MIN_EXPANSION_WEIGHT
        cutOffExpansions(lemmas, true);
        // restore verb form
        if (pos.equals("VBZ")) {
            // third person singular
            for (String exp : lemmas.keySet()) {
                double weight = lemmas.get(exp);
                String form = VerbFormConverter.infinitiveToThirdPersonS(exp);
                expansions.put(form, weight);
            }
        } else if (pos.equals("VBG")) {
            // gerund
            for (String exp : lemmas.keySet()) {
                double weight = lemmas.get(exp);
                String[] forms = VerbFormConverter.infinitiveToGerund(exp);
                for (String form : forms) expansions.put(form, weight);
            }
        } else if (pos.equals("VBD")) {
            // simple past
            for (String exp : lemmas.keySet()) {
                double weight = lemmas.get(exp);
                String[] forms = VerbFormConverter.infinitiveToSimplePast(exp);
                for (String form : forms) expansions.put(form, weight);
            }
        } else if (pos.equals("VBN")) {
            // past participle
            for (String exp : lemmas.keySet()) {
                double weight = lemmas.get(exp);
                String[] forms = VerbFormConverter.infinitiveToPastParticiple(exp);
                for (String form : forms) expansions.put(form, weight);
            }
        }
    } else if (pos.startsWith("JJ") || pos.startsWith("RB")) {
        // get modifier type
        POS modType = (pos.startsWith("JJ")) ? POS.ADJECTIVE : POS.ADVERB;
        // lemmatize adjectives and adverbs that are in WordNet
        String lemma = WordNet.getLemma(text, modType);
        if (lemma == null)
            lemma = text;
        // expand modifier
        for (Ontology ontology : ontologies) {
            Map<String, Double> expanded = ontology.expandModifier(lemma, modType);
            lemmas.putAll(expanded);
        }
        // ensure that there are at most MAX_EXPANSIONS expansions with
        // weights of at least MIN_EXPANSION_WEIGHT
        cutOffExpansions(lemmas, true);
    } else {
        // lemmatize nouns that are in WordNet
        String lemma;
        if (pos.startsWith("COMPOUND"))
            // compound
            lemma = WordNet.getCompoundLemma(text, POS.NOUN);
        else
            // single token
            lemma = WordNet.getLemma(text, POS.NOUN);
        if (lemma == null)
            lemma = text;
        // expand entity
        for (Ontology ontology : ontologies) {
            Map<String, Double> expanded = ontology.expandEntity(lemma);
            lemmas.putAll(expanded);
        }
        // ensure that there are at most MAX_EXPANSIONS expansions with
        // weights of at least MIN_EXPANSION_WEIGHT
        cutOffExpansions(lemmas, true);
    // TODO restore plural forms if possible
    }
    term.setExpansionLemmas(lemmas);
    term.setExpansions((expansions.size() > 0) ? expansions : lemmas);
}

Example 5 with Ontology

use of info.ephyra.nlp.semantics.ontologies.Ontology in project lucida by claritylab.

the class OpenEphyraServer method initFactoid.

/**
	 * Initializes the pipeline for factoid questions.
	 */
protected void initFactoid() {
    // question analysis
    Ontology wordNet = new WordNet();
    // - dictionaries for term extraction
    QuestionAnalysis.clearDictionaries();
    QuestionAnalysis.addDictionary(wordNet);
    // - ontologies for term expansion
    QuestionAnalysis.clearOntologies();
    QuestionAnalysis.addOntology(wordNet);
    // query generation
    QueryGeneration.clearQueryGenerators();
    QueryGeneration.addQueryGenerator(new BagOfWordsG());
    QueryGeneration.addQueryGenerator(new BagOfTermsG());
    QueryGeneration.addQueryGenerator(new PredicateG());
    QueryGeneration.addQueryGenerator(new QuestionInterpretationG());
    QueryGeneration.addQueryGenerator(new QuestionReformulationG());
    // search
    // - knowledge miners for unstructured knowledge sources
    Search.clearKnowledgeMiners();
    for (String[] indriIndices : IndriKM.getIndriIndices()) Search.addKnowledgeMiner(new IndriKM(indriIndices, false));
    // - knowledge annotators for (semi-)structured knowledge sources
    Search.clearKnowledgeAnnotators();
    /* Search.addKnowledgeAnnotator(new WikipediaKA("list.txt")); */
    // answer extraction and selection
    // (the filters are applied in this order)
    AnswerSelection.clearFilters();
    // - answer extraction filters
    AnswerSelection.addFilter(new AnswerTypeFilter());
    AnswerSelection.addFilter(new AnswerPatternFilter());
    AnswerSelection.addFilter(new PredicateExtractionFilter());
    AnswerSelection.addFilter(new FactoidsFromPredicatesFilter());
    AnswerSelection.addFilter(new TruncationFilter());
    // - answer selection filters
    AnswerSelection.addFilter(new StopwordFilter());
    AnswerSelection.addFilter(new QuestionKeywordsFilter());
    AnswerSelection.addFilter(new ScoreNormalizationFilter(NORMALIZER));
    AnswerSelection.addFilter(new ScoreCombinationFilter());
    AnswerSelection.addFilter(new FactoidSubsetFilter());
    AnswerSelection.addFilter(new DuplicateFilter());
    AnswerSelection.addFilter(new ScoreSorterFilter());
}