// Constructors ---------------------
        public EnglishTreebankParser(string dataDirectory, bool useTagDictionary, bool useCaseSensitiveTagDictionary, int beamSize, double advancePercentage)
        {
            var buildModelReader = new SharpEntropy.IO.BinaryGisModelReader(dataDirectory + "parser\\build.nbin");
            var buildModel = new SharpEntropy.GisModel(buildModelReader);

            var checkModelReader = new SharpEntropy.IO.BinaryGisModelReader(dataDirectory + "parser\\check.nbin");
            SharpEntropy.IMaximumEntropyModel checkModel = new SharpEntropy.GisModel(checkModelReader);

            EnglishTreebankPosTagger posTagger = useTagDictionary ?
                new EnglishTreebankPosTagger(dataDirectory + "parser\\tag.nbin", dataDirectory + "parser\\tagdict", useCaseSensitiveTagDictionary)
                : new EnglishTreebankPosTagger(dataDirectory + "parser\\tag.nbin");

            var chunker = new EnglishTreebankParserChunker(dataDirectory + "parser\\chunk.nbin");
            var headRules = new EnglishHeadRules(dataDirectory + "parser\\head_rules");

            _parser = new MaximumEntropyParser(buildModel, checkModel, posTagger, chunker, headRules, beamSize, advancePercentage);

            _tokenizer = new Tokenize.EnglishMaximumEntropyTokenizer(dataDirectory + "EnglishTok.nbin");
        }
Esempio n. 2
0
        // Constructors ---------------------

        public EnglishTreebankParser(string dataDirectory, bool useTagDictionary, bool useCaseSensitiveTagDictionary, int beamSize, double advancePercentage)
        {
            var buildModelReader = new SharpEntropy.IO.BinaryGisModelReader(dataDirectory + "parser\\build.nbin");
            var buildModel       = new SharpEntropy.GisModel(buildModelReader);

            var checkModelReader = new SharpEntropy.IO.BinaryGisModelReader(dataDirectory + "parser\\check.nbin");

            SharpEntropy.IMaximumEntropyModel checkModel = new SharpEntropy.GisModel(checkModelReader);

            EnglishTreebankPosTagger posTagger = useTagDictionary ?
                                                 new EnglishTreebankPosTagger(dataDirectory + "parser\\tag.nbin", dataDirectory + "parser\\tagdict", useCaseSensitiveTagDictionary)
                        : new EnglishTreebankPosTagger(dataDirectory + "parser\\tag.nbin");

            var chunker   = new EnglishTreebankParserChunker(dataDirectory + "parser\\chunk.nbin");
            var headRules = new EnglishHeadRules(dataDirectory + "parser\\head_rules");

            _parser = new MaximumEntropyParser(buildModel, checkModel, posTagger, chunker, headRules, beamSize, advancePercentage);

            _tokenizer = new Tokenize.EnglishMaximumEntropyTokenizer(dataDirectory + "EnglishTok.nbin");
        }