edu.illinois.cs.cogcomp.wsim.wordnet

Class Paraphrase

java.lang.Object

edu.illinois.cs.cogcomp.sim.StringMetric<edu.illinois.cs.cogcomp.sim.MetricWord>

edu.illinois.cs.cogcomp.wsim.wordnet.Paraphrase

All Implemented Interfaces:

Metric<edu.illinois.cs.cogcomp.sim.MetricWord>

public class Paraphrase
extends StringMetric<edu.illinois.cs.cogcomp.sim.MetricWord>

Class to compute paraphrase based similarity

Author:

John Wieting, mssammon, ngupta18, sgupta96

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	Paraphrase.Para

Field Summary

Fields

Modifier and Type	Field and Description
HashMap<String,Paraphrase.Para>	paraphrase_map_nopos
HashMap<String,Paraphrase.Para>	paraphrase_map_pos

Constructor Summary

Constructors

Constructor and Description
Paraphrase() default constructor: expects all resources to be defaults, and will look for them on classpath
Paraphrase(ResourceManager rm) Constructor

Method Summary

Modifier and Type	Method and Description
MetricResponse	compare(edu.illinois.cs.cogcomp.sim.MetricWord arg1, edu.illinois.cs.cogcomp.sim.MetricWord arg2) Calculates similarity between word1 and word2 TODO: determine catalog of POS values for POS tags of word1 and word2
MetricResponse	compareString(String arg1, String arg2) Calculates similarity between two words when only word strings are present.
String	getName() returns the name of this metric.
static String	getPOS(String pos) TODO: use enum type for valid POS
static void	main(String[] args) Contains sample invocations
protected edu.illinois.cs.cogcomp.sim.MetricWord	wrapStringArgument(String word) construct a T instance from just a String to allow the StringMetric.compareStringValues(java.util.Map<java.lang.String, java.lang.String>) to interact with .compare

Methods inherited from class edu.illinois.cs.cogcomp.sim.StringMetric

compareStringValues

Methods inherited from class java.lang.Object

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

Field Detail

paraphrase_map_pos

public HashMap<String,Paraphrase.Para> paraphrase_map_pos

paraphrase_map_nopos

public HashMap<String,Paraphrase.Para> paraphrase_map_nopos

Constructor Detail

Paraphrase

public Paraphrase()
           throws IOException

default constructor: expects all resources to be defaults, and will look for them on classpath

Throws:

IOException

Paraphrase

public Paraphrase(ResourceManager rm)
           throws IOException

Constructor

Parameters:

rm - ResourceManager with non-default properties set

Throws:

IOException

Method Detail

main

public static void main(String[] args)
                 throws IOException

Contains sample invocations

Parameters:

args - optional: config file containing path to the paraphrase resource, as in config/sampleConfig.txt

Throws:

IOException

getPOS

public static String getPOS(String pos)

TODO: use enum type for valid POS

Parameters:

pos - Part of Speech Tag

Returns:

some simpler value of the pos tag

getName

public String getName()

returns the name of this metric. Used as the reason by the default StringMetric.compareStringValues(java.util.Map<java.lang.String, java.lang.String>) method.

Specified by:

getName in class StringMetric<edu.illinois.cs.cogcomp.sim.MetricWord>

Returns:

the name of this metric.

wrapStringArgument

protected edu.illinois.cs.cogcomp.sim.MetricWord wrapStringArgument(String word)

construct a T instance from just a String to allow the StringMetric.compareStringValues(java.util.Map<java.lang.String, java.lang.String>) to interact with .compare

Specified by:

wrapStringArgument in class StringMetric<edu.illinois.cs.cogcomp.sim.MetricWord>

Parameters:

word - the word to wrap

Returns:

the corresponding object of type T specified by the implementor

compare

public MetricResponse compare(edu.illinois.cs.cogcomp.sim.MetricWord arg1,
                              edu.illinois.cs.cogcomp.sim.MetricWord arg2)
                       throws IllegalArgumentException,
                              org.apache.commons.lang.NotImplementedException

Calculates similarity between word1 and word2 TODO: determine catalog of POS values for POS tags of word1 and word2

Parameters:

arg1 - 1st component (word + POS tag)

arg2 - 2nd component

Returns:

similarity score between words plus a "reason"

Throws:

IllegalArgumentException

org.apache.commons.lang.NotImplementedException

compareString

public MetricResponse compareString(String arg1,
                                    String arg2)
                             throws IllegalArgumentException

Calculates similarity between two words when only word strings are present. Metric is Asymmetric

Parameters:

arg1 - 1st component

arg2 - 2nd component

Returns:

similarity score between word1 and word2 plus a "reason"

Throws:

IllegalArgumentException