edu.illinois.cs.cogcomp.edison.utilities

Class POSMikheevCounter

java.lang.Object

edu.illinois.cs.cogcomp.edison.utilities.POSBaseLineCounter

edu.illinois.cs.cogcomp.edison.utilities.POSMikheevCounter

public class POSMikheevCounter
extends POSBaseLineCounter

A poor man's Mikheev rules implementation, in that there is no morphology. Only rules involving all-letter suffixes of lengths 3 and 4 of words whose length is at least 2 greater than the suffix are extracted. The learned rules are divided into three categories: words that are capitalized and appear first in the sentence, words that are capitalized and don't appear first in the sentence, and uncapitalized, unhyphenated words.

Author:

Xinbo Wu

Field Summary

Fields

Modifier and Type	Field and Description
protected HashMap<String,TreeMap<String,Integer>>	firstCapitalized A map for capitalized words appearing first in the sentence.
protected HashMap<String,TreeMap<String,Integer>>	notFirstCapitalized A map for capitalized words not appearing first in the sentence.

Fields inherited from class edu.illinois.cs.cogcomp.edison.utilities.POSBaseLineCounter

corpusName, table

Constructor Summary

Constructors

Constructor and Description
POSMikheevCounter(String corpusName)

Method Summary

Modifier and Type	Method and Description
Set<String>	allowableTags(TextAnnotation ta, int tokenId) Returns the set of tags that the given word's suffix has been observed with, or a reasonable default if the suffix has never been observed.
void	buildTable(String home) A table is built from either a given source corpus file or source corpus directory by counting the number of times that each suffix-POS association in a source corpus.
void	count(HashMap<String,TreeMap<String,Integer>> table, String suffix, String tag) Increments the count for the given suffix and tag.
void	doneLearning() Runs after all learning is complete.
void	forget() Clears out the table to start fresh.
void	prune(HashMap<String,TreeMap<String,Integer>> table) Prunes the specified table.
static POSMikheevCounter	read(String json) Read the an instance of POSMikheevCounter class from JSON format
String	tag(int tokenId, TextAnnotation ta) Give a tag for a given form
static String	write(POSMikheevCounter counter) Write an instance of POSMikheevCounter class to JSON format

Methods inherited from class edu.illinois.cs.cogcomp.edison.utilities.POSBaseLineCounter

allowableTags, count, getCorpusName, looksLikeNumber, observed, observedCount, write

Methods inherited from class java.lang.Object

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

Field Detail

firstCapitalized

protected HashMap<String,TreeMap<String,Integer>> firstCapitalized

A map for capitalized words appearing first in the sentence.

notFirstCapitalized

protected HashMap<String,TreeMap<String,Integer>> notFirstCapitalized

A map for capitalized words not appearing first in the sentence.

Constructor Detail

POSMikheevCounter

public POSMikheevCounter(String corpusName)

Method Detail

buildTable

public void buildTable(String home)
                throws Exception

A table is built from either a given source corpus file or source corpus directory by counting the number of times that each suffix-POS association in a source corpus.

Overrides:

buildTable in class POSBaseLineCounter

Parameters:

home - file name or directory name of the source corpus

Throws:

Exception

count

public void count(HashMap<String,TreeMap<String,Integer>> table,
                  String suffix,
                  String tag)

Increments the count for the given suffix and tag.

Parameters:

table - The table in which a count should be incremented.

suffix - The suffix.

tag - The POS tag.

doneLearning

public void doneLearning()

Runs after all learning is complete.

prune

public void prune(HashMap<String,TreeMap<String,Integer>> table)

Prunes the specified table.

Parameters:

table - The table.

forget

public void forget()

Clears out the table to start fresh.

Overrides:

forget in class POSBaseLineCounter

tag

public String tag(int tokenId,
                  TextAnnotation ta)

Give a tag for a given form

Overrides:

tag in class POSBaseLineCounter

allowableTags

public Set<String> allowableTags(TextAnnotation ta,
                                 int tokenId)

Returns the set of tags that the given word's suffix has been observed with, or a reasonable default if the suffix has never been observed.

Returns:

The set of tags observed in association with the given word's suffix, or a reasonable default if the suffix has never been observed.

write

public static String write(POSMikheevCounter counter)

Write an instance of POSMikheevCounter class to JSON format

read

public static POSMikheevCounter read(String json)

Read the an instance of POSMikheevCounter class from JSON format