edu.illinois.cs.cogcomp.edison.features.helpers

Class ParseHelper

java.lang.Object

edu.illinois.cs.cogcomp.edison.features.helpers.ParseHelper

public class ParseHelper
extends Object

This class contains helper functions for retrieving parse trees and related features from the parse view.

Author:

Vivek Srikumar

Sep 30, 2009

Field Summary

Fields

Modifier and Type	Field and Description
static String	PATH_DOWN_STRING This string indicates a path going down a parse tree in the path string.
static String	PATH_UP_STRING This string indicates a path going up a parse tree in the path string.

Constructor Summary

Constructors

Constructor and Description
ParseHelper()

Method Summary

Modifier and Type	Method and Description
static String[]	getAllPhraseSiblingLabels(String parseViewName, Constituent constituent) Get the labels of all the siblings of the parse tree node that covers the input constituent.
static String[]	getAllSiblingLabels(Tree<String> node) Get the labels of all siblings of a given tree node.
static <T> Tree<T>	getCommonAncestor(Tree<T> t1, Tree<T> t2, Tree<T> tree) Deprecated.
static int	getHeadWordPosition(Constituent c, HeadFinderBase headFinder, String parseViewName) Get the head word of a constituent using the HeadFinderBase that is passed as an argument.
static Tree<String>	getParseTree(String parseViewName, Sentence s) Get the parse tree of a sentence.
static Tree<String>	getParseTree(String parseViewName, TextAnnotation ta, int sentenceId) Get the parse tree of the sentenceIdth sentence from the text annotation.
static Tree<String>	getParseTree(TextAnnotation ta, int sentenceId) Deprecated.
static Tree<String>	getParseTreeCovering(String parseViewName, Constituent c) Get a parse tree from a text annotation that covers the specified constituent.
static <T> Pair<List<Tree<T>>,List<Tree<T>>>	getPath(Tree<T> start, Tree<T> end, Tree<T> tree, int maxDepth) Deprecated.
static <T> String	getPathString(Tree<T> start, Tree<T> end, Tree<T> tree, int maxDepth) Deprecated.
static <T> String	getPathStringIgnoreLexicalItems(Tree<T> start, Tree<T> end, Tree<T> tree, int maxDepth) Deprecated.
static <T> String	getPathStringToCommonAncestor(Tree<T> start, Tree<T> end, Tree<T> tree, int maxDepth) Deprecated.
static <T> List<T>	getPathToRoot(Tree<T> tree, Tree<T> leaf, int maxDepth) Deprecated.
static <T> List<Tree<T>>	getPathTreesToRoot(Tree<T> tree, Tree<T> node, int maxDepth) Deprecated.
static Constituent	getPhraseFromHead(Constituent predicate, Constituent argHead, String parseViewName) Primarily a fix for prepSRL objects; converts them from single head words to constituents.
static String	getSubcatFrame(Tree<String> yieldNode) Assuming that the tree comes with lexical items and POS tags, the subcat frame for the verb can be found by going to the parent of the POS tag (which is probably a VP) and listing its children.
static String	getTerminalString(Tree<String> tree)
static String	getTerminalStringSentence(Tree<String> tree)
static List<String>	getTerminalTokens(Tree<String> tree)
static Tree<Pair<String,IntPair>>	getTokenIndexedCleanedParseTreeNodeCovering(Constituent c, String parseViewName)
static Tree<Pair<String,IntPair>>	getTokenIndexedParseTreeNodeCovering(String parseViewName, Constituent c)
static Tree<Pair<String,IntPair>>	getTokenIndexedTreeCovering(Tree<String> parse, int start, int end) From a parse tree and a span that is specified with the start and end (exclusive), this function returns a tree that corresponds to the subtree that covers the span.
static Tree<String>	getTreeCovering(Tree<String> parse, int start, int end)

Methods inherited from class java.lang.Object

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

Field Detail

PATH_UP_STRING

public static final String PATH_UP_STRING

This string indicates a path going up a parse tree in the path string.

See Also:

Constant Field Values

PATH_DOWN_STRING

public static final String PATH_DOWN_STRING

This string indicates a path going down a parse tree in the path string.

See Also:

Constant Field Values

Constructor Detail

ParseHelper

public ParseHelper()

Method Detail

getAllPhraseSiblingLabels

public static String[] getAllPhraseSiblingLabels(String parseViewName,
                                                 Constituent constituent)

Get the labels of all the siblings of the parse tree node that covers the input constituent. Note that the input constituent could be from any view. The output includes the phrase label of this constituent too.

Parameters:

parseViewName - The name of the parse view. This might typically be one of ViewNames.PARSE_CHARNIAK or ViewNames.PARSE_STANFORD

constituent - The constituent whose sibling phrases are required.

Returns:

An array of strings. If the input constituent is the entire sentence, then an array of size zero is returned to indicate that the corresponding parse tree node does not have any siblings.

getAllSiblingLabels

public static String[] getAllSiblingLabels(Tree<String> node)

Get the labels of all siblings of a given tree node.

Parameters:

node - The node whose siblings are required.

Returns:

The labels of the siblings of the input node, as an array of strings. If the input is the root of the tree, then a string array of size zero is returned to indicate that the corresponding parse tree node does not have any siblings.

getCommonAncestor

@Deprecated
public static <T> Tree<T> getCommonAncestor(Tree<T> t1,
                                                        Tree<T> t2,
                                                        Tree<T> tree)
                                                 throws Exception

Deprecated.

Finds a common ancestor of two trees which are both contained in a larger tree. This searches the path from the head of the trees to the root of the trees till it finds a common node. If none is found, then the function returns null.

Type Parameters:

T - The type of tree

Parameters:

t1 - The first tree

t2 - The second tree

tree - The tree that contains t1 and t2.

Returns:

The tree that is the common ancestor of t1 and t2 . If none is found, then the function returns null.

Throws:

Exception

getParseTree

public static Tree<String> getParseTree(String parseViewName,
                                        Sentence s)

Get the parse tree of a sentence. This code assumes that the view called parseViewName exists in the text annotation.

Parameters:

parseViewName - The name of the parse view

s - The sentence

Returns:

The parse tree of the sentence

getParseTree

public static Tree<String> getParseTree(String parseViewName,
                                        TextAnnotation ta,
                                        int sentenceId)

Get the parse tree of the sentenceIdth sentence from the text annotation. This code assumes that the view called parseViewName exists in the text annotation.

Parameters:

parseViewName - The name of the parse view

ta - The text annotation object

sentenceId - The sentence whose parse tree is required

Returns:

The parse tree of the sentenceIdth sentence

getParseTree

@Deprecated
public static Tree<String> getParseTree(TextAnnotation ta,
                                                    int sentenceId)

Deprecated.

Get the Charniak parse tree of the sentenceIdth sentence from the text annotation. This code assumes that the view called ViewNames.PARSE_CHARNIAK exists in the text annotation.

Parameters:

ta - The text annotation object

sentenceId - The sentence whose parse tree is required

Returns:

The parse tree of the sentenceIdth sentence

getParseTreeCovering

public static Tree<String> getParseTreeCovering(String parseViewName,
                                                Constituent c)

Get a parse tree from a text annotation that covers the specified constituent.

Parameters:

parseViewName - The name of the parse view

c - The constituent that we care about

Returns:

The portion of the parse tree of the TextAnnotation to which the constituent belongs which covers the constituent.

getPath

@Deprecated
public static <T> Pair<List<Tree<T>>,List<Tree<T>>> getPath(Tree<T> start,
                                                                        Tree<T> end,
                                                                        Tree<T> tree,
                                                                        int maxDepth)
                                                                 throws Exception

Deprecated.

Returns a pair of paths. The first element of the pair is the path up from the start node to the common ancestor of start and end. The second element is the path down from the common ancestor to the end node.

Throws:

Exception

getPathString

@Deprecated
public static <T> String getPathString(Tree<T> start,
                                                   Tree<T> end,
                                                   Tree<T> tree,
                                                   int maxDepth)
                                            throws Exception

Deprecated.

Get a string representing the path between trees start and end that belong to the tree tree

Throws:

Exception

getPathStringIgnoreLexicalItems

@Deprecated
public static <T> String getPathStringIgnoreLexicalItems(Tree<T> start,
                                                                     Tree<T> end,
                                                                     Tree<T> tree,
                                                                     int maxDepth)
                                                              throws Exception

Deprecated.

Throws:

Exception

getPathStringToCommonAncestor

@Deprecated
public static <T> String getPathStringToCommonAncestor(Tree<T> start,
                                                                   Tree<T> end,
                                                                   Tree<T> tree,
                                                                   int maxDepth)
                                                            throws Exception

Deprecated.

Throws:

Exception

getPathToRoot

@Deprecated
public static <T> List<T> getPathToRoot(Tree<T> tree,
                                                    Tree<T> leaf,
                                                    int maxDepth)
                                             throws Exception

Deprecated.

Throws:

Exception

getPathTreesToRoot

@Deprecated
public static <T> List<Tree<T>> getPathTreesToRoot(Tree<T> tree,
                                                               Tree<T> node,
                                                               int maxDepth)
                                                        throws Exception

Deprecated.

This function returns a list of trees from a node in the tree to the root. If the node is deeper than maxdepth, then the bottom maxDepth nodes are returned.

Throws:

Exception

getSubcatFrame

public static String getSubcatFrame(Tree<String> yieldNode)

Assuming that the tree comes with lexical items and POS tags, the subcat frame for the verb can be found by going to the parent of the POS tag (which is probably a VP) and listing its children.

getTerminalString

public static String getTerminalString(Tree<String> tree)

getTerminalStringSentence

public static String getTerminalStringSentence(Tree<String> tree)

getTerminalTokens

public static List<String> getTerminalTokens(Tree<String> tree)

getTokenIndexedCleanedParseTreeNodeCovering

public static Tree<Pair<String,IntPair>> getTokenIndexedCleanedParseTreeNodeCovering(Constituent c,
                                                                                     String parseViewName)

getTokenIndexedParseTreeNodeCovering

public static Tree<Pair<String,IntPair>> getTokenIndexedParseTreeNodeCovering(String parseViewName,
                                                                              Constituent c)

getTokenIndexedTreeCovering

public static Tree<Pair<String,IntPair>> getTokenIndexedTreeCovering(Tree<String> parse,
                                                                     int start,
                                                                     int end)

From a parse tree and a span that is specified with the start and end (exclusive), this function returns a tree that corresponds to the subtree that covers the span. Each node in the new tree corresponds to a node in the input tree and is labeled with the label of the original node along with the span that this node covered in the original tree.

Returns:

A new tree that covers the specified span and each node specifies the label and the span of the original tree that it covers.

getTreeCovering

public static Tree<String> getTreeCovering(Tree<String> parse,
                                           int start,
                                           int end)

getHeadWordPosition

public static int getHeadWordPosition(Constituent c,
                                      HeadFinderBase headFinder,
                                      String parseViewName)
                               throws Exception

Get the head word of a constituent using the HeadFinderBase that is passed as an argument. To use this function, first, a head finder should be created. For example:

   TextAnnotation ta = ... // some text annotation
   Constituent c = ... // some constituent
 
   CollinsHeadFinder headFinder = new CollinsHeadFinder();
 
   int headId = ParseHelper.getHeadWordPosition(c, headFinder);
 
   // now we can do other things with the headId.
   String headWord = WordHelpers.getWord(ta, headId);
 

Parameters:

parseViewName - The name of the view which contains the parse trees

c - The constituent whose head we wish to find

headFinder - The head finder

Returns:

The index of the head word. The word itself can be obtained later by calling WordHelpers.getWord(TextAnnotation, int).

Throws:

EdisonException

Exception

getPhraseFromHead

public static Constituent getPhraseFromHead(Constituent predicate,
                                            Constituent argHead,
                                            String parseViewName)

Primarily a fix for prepSRL objects; converts them from single head words to constituents. E.g. for the sentence "the man with the telescope", the object of the preposition will be "the telescope" instead of just "telescope".

Parameters:

predicate - The predicate of the construction (e.g. "with")

argHead - The head-word of the argument of the construction (e.g. "telescope")

parseViewName - The name of the parse view used to extract the phrase-structure tree

Returns:

The full constituent phrase containing the argument head