edu.ucla.sspace.dependency

Class CoNLLDependencyExtractor

java.lang.Object

edu.ucla.sspace.dependency.CoNLLDependencyExtractor

All Implemented Interfaces:

DependencyExtractor

Direct Known Subclasses:

WaCKyDependencyExtractor

public class CoNLLDependencyExtractor
extends Object
implements DependencyExtractor

A class for extracting dependency parsed sentences in the CoNLL format, which are generated by the Malt Parser. The ordering of the CoNLL format can be specified with an xml file following the format specified here. This configuration file simply specifies the ordering of the dependency features. By default, the extractor assumes the default format used by the Malt Parser

Parsed sentences are returned as an array of DependencyTreeNode innstances. The nodes contain relations between each word in the sentence. The nodes in the returned array are ordered by the ordering of word occurrences.

This class optionally supports filtering sentences to remove words. The nodes for those removed words will still remain in the parse tree. Similarly, the relations connecting the removed words will also existing. However, the DependencyTreeNode.word() method will return IteratorFactory.EMPTY_TOKEN to indicate that the node's text was filtered out. Note that the node will still have the original part of speech.

Author:

Keith Stevens

Constructor Summary

Constructors

Constructor and Description
CoNLLDependencyExtractor() Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses.
CoNLLDependencyExtractor(String configFile) Creates a new CoNLLDependencyExtractor by parsing a Malt configuration file, which specifies the order in which the output is formatted.
CoNLLDependencyExtractor(String configFile, TokenFilter filter, Stemmer stemmer) Creates a new CoNLLDependencyExtractor by parsing a Malt configuration file, which specifies the order in which the output is formatted and uses the given TokenFilter and Stemmer.
CoNLLDependencyExtractor(TokenFilter filter, Stemmer stemmer) Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses and uses the given TokenFilter and Stemmer.
CoNLLDependencyExtractor(TokenFilter filter, Stemmer stemmer, int idIndex, int formIndex, int lemmaIndex, int posIndex, int parentIndex, int relationIndex) Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses and uses the given TokenFilter and Stemmer and the given indices for each feature.

Method Summary

Methods

Modifier and Type	Method and Description
DependencyTreeNode[]	readNextTree(BufferedReader reader) Extracts a dependency parse tree from the provided reader.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

CoNLLDependencyExtractor

public CoNLLDependencyExtractor()

Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses.

CoNLLDependencyExtractor

public CoNLLDependencyExtractor(TokenFilter filter,
                        Stemmer stemmer)

Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses and uses the given TokenFilter and Stemmer.

CoNLLDependencyExtractor

public CoNLLDependencyExtractor(TokenFilter filter,
                        Stemmer stemmer,
                        int idIndex,
                        int formIndex,
                        int lemmaIndex,
                        int posIndex,
                        int parentIndex,
                        int relationIndex)

Creates a new CoNLLDependencyExtractor that assumes the default ordering for Malt dependency parses and uses the given TokenFilter and Stemmer and the given indices for each feature.

CoNLLDependencyExtractor

public CoNLLDependencyExtractor(String configFile)

Creates a new CoNLLDependencyExtractor by parsing a Malt configuration file, which specifies the order in which the output is formatted.

CoNLLDependencyExtractor

public CoNLLDependencyExtractor(String configFile,
                        TokenFilter filter,
                        Stemmer stemmer)

Creates a new CoNLLDependencyExtractor by parsing a Malt configuration file, which specifies the order in which the output is formatted and uses the given TokenFilter and Stemmer.

Method Detail

readNextTree

public DependencyTreeNode[] readNextTree(BufferedReader reader)
                                  throws IOException

Extracts a dependency parse tree from the provided reader. The tree is assumed to be in the CoNLL format.

The CoNNLL features that are of interest are ID, LEMMA or FORM, POSTAG, HEAD, and DEPREL which are the id of the node, the string for the word at this position, the part of speech tag, the parent of this node, and the relation to the parent, respectively. These features will be extracted and returned as an array based tree structure ordered by the occurrence of the words in the sentence.

Specified by:

readNextTree in interface DependencyExtractor

Parameters:

reader - a reader containing one or more parse trees in the CoNLL format

Returns:

an array of DependencyTreeNodes that compose a dependency tree, or null if no tree is present in the reader.

Throws:

IOException - when errors are encountered during reading