org.ow2.weblab.service.language
Class NGramProfilesPatched

java.lang.Object
  org.ow2.weblab.service.language.NGramProfilesPatched

public class NGramProfilesPatched
extends java.lang.Object

This is a patched version of NGramProfiles class from CNGram project. It has been added to enable the use of custom language models. When using the default version, it's only possible to load languages models that are present in the jar, next to the NGramProfiles class. We only copy-paste the code, add the right behaviour by adding a constructor and an init method on a folder (containing language models). We also removed some warning by changing the access right on some fields and qualifying accesses with this. We also handle generics when possible. And last but not least, we change the code of original loading method to use the same ClassLoader than before (i.e. original NGramProfiles one). "Manage a set of profiles and determine "most similar" ones to a given profile. Allows access to the complete results of previous last ranking. Note this uses a competetive ranking approach, which is memory efficient, time efficient for not too many languages and provides contextual scoring of ngrams."

Author:

Inspired from frank nestel, EADS IPCC Team

See Also:

NGramProfiles

Nested Class Summary

static interface	NGramProfilesPatched.Ranker
static interface	NGramProfilesPatched.RankResult

Field Summary

static java.text.DecimalFormat	DF
static char	END_CHAR
static double	LOWSTATSAFETY
static java.lang.String	NOLANGNAME

Constructor Summary

NGramProfilesPatched()

NGramProfilesPatched(java.io.BufferedReader br)

NGramProfilesPatched(java.io.File folder)

NGramProfilesPatched(int mode)

Method Summary

static char	charAt(java.lang.CharSequence cs, int pos)
java.util.Set<de.spieleck.app.cngram.NGram>	getAllNGrams()
int	getProfileCount()
java.lang.String	getProfileName(int i)
NGramProfilesPatched.Ranker	getRanker()
protected void	init(java.io.File folder)
NGramProfilesPatched.RankResult	rank(de.spieleck.app.cngram.NGramMetric metric, de.spieleck.app.cngram.NGramProfile profile) Note this class returns a complete match result, for the sake of thread safety!

Methods inherited from class java.lang.Object

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

Field Detail

NOLANGNAME

public static final java.lang.String NOLANGNAME

See Also:

Constant Field Values

END_CHAR

public static final char END_CHAR

See Also:

Constant Field Values

DF

public static final java.text.DecimalFormat DF

LOWSTATSAFETY

public static final double LOWSTATSAFETY

See Also:

Constant Field Values

Constructor Detail

NGramProfilesPatched

public NGramProfilesPatched()
                     throws java.io.IOException

Throws:

java.io.IOException

NGramProfilesPatched

public NGramProfilesPatched(int mode)
                     throws java.io.IOException

Throws:

java.io.IOException

NGramProfilesPatched

public NGramProfilesPatched(java.io.BufferedReader br)
                     throws java.io.IOException

Throws:

java.io.IOException

NGramProfilesPatched

public NGramProfilesPatched(java.io.File folder)
                     throws java.io.IOException

Throws:

java.io.IOException

Method Detail

getRanker

public NGramProfilesPatched.Ranker getRanker()

getProfileName

public java.lang.String getProfileName(int i)

charAt

public static final char charAt(java.lang.CharSequence cs,
                                int pos)

rank

public NGramProfilesPatched.RankResult rank(de.spieleck.app.cngram.NGramMetric metric,
                                            de.spieleck.app.cngram.NGramProfile profile)

Note this class returns a complete match result, for the sake of thread safety!

getProfileCount

public int getProfileCount()

getAllNGrams

public java.util.Set<de.spieleck.app.cngram.NGram> getAllNGrams()

init

protected void init(java.io.File folder)
             throws java.io.IOException

Throws:

java.io.IOException