info.ephyra.questionanalysis
Class QuestionPattern

java.lang.Object
  info.ephyra.questionanalysis.QuestionPattern

public class QuestionPattern
extends java.lang.Object

A QuestionPattern is applied to a question to determine the TARGET of the question, CONTEXT information and the PROPERTY the question asks for.

Version:

2006-04-20

Author:

Nico Schlaefer

Field Summary

private int[]	contextIDs IDs of 0 to n groups that represent the CONTEXT of the question.
private static int	MAX_CONTEXT Maximum length of a CONTEXT object in tokens.
private static int	MAX_TARGET Maximum length of a TARGET object in tokens.
private java.util.regex.Pattern	pattern The Pattern that is applied to a question string.
private java.lang.String	property The PROPERTY that a question which matches this pattern asks for.
private int	targetID ID of the group that represents the TARGET of the question.

Constructor Summary

QuestionPattern(java.lang.String expr, java.lang.String prop)
Creates a QuestionPattern from a descriptor that is a regular expression but additionally contains the following tags: <TO> - exactly one TARGET tag <CO> - an arbitrary number of CONTEXT tags

Method Summary

QuestionInterpretation	apply(java.lang.String qn, java.lang.String stemmed) Applies the pattern to a question.
private QuestionInterpretation	ensureNounPhrases(java.lang.String qn, QuestionInterpretation qi) Ensures that the TARGET and CONTEXT objects are noun phrases and splits the objects along prepositions and punctuation marks.
private java.lang.String	formatObject(java.lang.String object) Formats the extracted TARGET and CONTEXT objects.
java.lang.String	getProperty() Returns the PROPERTY that a question which matches the pattern asks for.
private java.lang.String	replaceContextTags(java.lang.String expr) Sets the contextIDs field and replaces the CONTEXT tags by capturing groups.
private java.lang.String	replaceTargetTag(java.lang.String expr) Sets the targetID field and replaces the TARGET tag by a capturing group.

Methods inherited from class java.lang.Object

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

Field Detail

MAX_TARGET

private static final int MAX_TARGET

Maximum length of a TARGET object in tokens.

See Also:

Constant Field Values

MAX_CONTEXT

private static final int MAX_CONTEXT

Maximum length of a CONTEXT object in tokens.

See Also:

Constant Field Values

pattern

private java.util.regex.Pattern pattern

The Pattern that is applied to a question string.

property

private java.lang.String property

The PROPERTY that a question which matches this pattern asks for.

targetID

private int targetID

ID of the group that represents the TARGET of the question.

contextIDs

private int[] contextIDs

IDs of 0 to n groups that represent the CONTEXT of the question.

Constructor Detail

QuestionPattern

public QuestionPattern(java.lang.String expr,
                       java.lang.String prop)

Creates a QuestionPattern from a descriptor that is a regular expression but additionally contains the following tags:

• <TO> - exactly one TARGET tag
• <CO> - an arbitrary number of CONTEXT tags

Parameters:

expr - pattern descriptor

prop - PROPERTY that a question which matches the pattern asks for

Method Detail

replaceTargetTag

private java.lang.String replaceTargetTag(java.lang.String expr)

Sets the targetID field and replaces the TARGET tag by a capturing group.

Parameters:

expr - pattern descriptor

Returns:

descriptor without TARGET tag

replaceContextTags

private java.lang.String replaceContextTags(java.lang.String expr)

Sets the contextIDs field and replaces the CONTEXT tags by capturing groups.

Parameters:

expr - pattern descriptor

Returns:

descriptor without CONTEXT tags

ensureNounPhrases

private QuestionInterpretation ensureNounPhrases(java.lang.String qn,
                                                 QuestionInterpretation qi)

Ensures that the TARGET and CONTEXT objects are noun phrases and splits the objects along prepositions and punctuation marks.

Parameters:

qn - normalized question string

qi - question interpretation

Returns:

modified question interpretation

formatObject

private java.lang.String formatObject(java.lang.String object)

Formats the extracted TARGET and CONTEXT objects.

Parameters:

object - TARGET or CONTEXT object

Returns:

formatted object

getProperty

public java.lang.String getProperty()

Returns the PROPERTY that a question which matches the pattern asks for.

Returns:

the PROPERTY

apply

public QuestionInterpretation apply(java.lang.String qn,
                                    java.lang.String stemmed)

Applies the pattern to a question. If the question matches the pattern, a QuestionInterpretation is returned, else null.

Parameters:

qn - normalized question string

stemmed - stemmed question string

Returns:

interpretation of the question or null, if the question does not match the pattern