edu.illinois.cs.cogcomp.edison.features

Class FeatureUtilities

java.lang.Object

edu.illinois.cs.cogcomp.edison.features.FeatureUtilities

public class FeatureUtilities
extends Object

Author:

Vivek Srikumar

Constructor Summary

Constructors

Constructor and Description
FeatureUtilities()

Method Summary

Modifier and Type	Method and Description
static FeatureExtractor	conjoin(FeatureExtractor f1, FeatureExtractor f2) Conjoins two feature extractors.
static Set<Feature>	conjoin(Set<Feature> feats1, Set<Feature> feats2) Conjoins two feature sets.
static Feature	conjunct(String left, String right) Conjoins two discrete features corresponding to the strings left and right.
static Pair<int[],double[]>	convert(Set<Feature> features, edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon, boolean trainingMode) Convert a feature set into a pair of arrays of integers and doubles by looking up the feature name in the provided lexicon.
static Set<Feature>	deserializeFeatureSet(byte[] bytes) Deserializes a byte array into a set of features
static edu.illinois.cs.cogcomp.lbjava.classify.Feature	getFeature(edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon, int id) This function gets the feature from the lexicon using the id.
static int	getFeatureId(edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon, boolean trainingMode, Feature feature) Returns the id of the feature according to the lexicon.
static Map<String,Float>	getFeatureMap(FeatureExtractor fex, Constituent c) Create a feature map using the specified feature extractor on the input constituent.
static Set<Feature>	getFeatures(Counter<String> counter) Convert a counter over strings into a set of real valued features.
static Set<Feature>	getFeatures(Map<String,Float> features) Convert a map from feature names to values into a set of real valued features.
static Set<Feature>	getFeatures(Set<String> features) Convert a set of strings into a set of discrete features.
static Set<String>	getFeatureSet(FeatureExtractor fex, Constituent c) Create a feature set using the specified feature extractor on the input constituent.
static edu.illinois.cs.cogcomp.lbjava.classify.Feature	getLBJFeature(Feature feature) Convert an edison feature into an LBJ feature
static edu.illinois.cs.cogcomp.lbjava.classify.Classifier	getLBJFeatureExtractor(FeatureExtractor fex) Convert an edison feature extractor into an LBJ classifier
static edu.illinois.cs.cogcomp.lbjava.classify.FeatureVector	getLBJFeatures(Set<Feature> features) Converts a Set of features into an LBJ friendly feature vector.
static Set<Feature>	prefix(String prefix, Set<Feature> features) Add a prefix to all the features in the set.
static byte[]	serializeFeatureSet(Set<Feature> features) Serializes a set of features into a byte array.

Methods inherited from class java.lang.Object

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

Constructor Detail

FeatureUtilities

public FeatureUtilities()

Method Detail

getFeatures

public static Set<Feature> getFeatures(Set<String> features)

Convert a set of strings into a set of discrete features.

getFeatures

public static Set<Feature> getFeatures(Map<String,Float> features)

Convert a map from feature names to values into a set of real valued features.

getFeatures

public static Set<Feature> getFeatures(Counter<String> counter)

Convert a counter over strings into a set of real valued features.

prefix

public static Set<Feature> prefix(String prefix,
                                  Set<Feature> features)

Add a prefix to all the features in the set.

Parameters:

prefix - The prefix to be added.

features - The feature set

getLBJFeatures

public static edu.illinois.cs.cogcomp.lbjava.classify.FeatureVector getLBJFeatures(Set<Feature> features)

Converts a Set of features into an LBJ friendly feature vector.

getLBJFeature

public static edu.illinois.cs.cogcomp.lbjava.classify.Feature getLBJFeature(Feature feature)

Convert an edison feature into an LBJ feature

getLBJFeatureExtractor

public static edu.illinois.cs.cogcomp.lbjava.classify.Classifier getLBJFeatureExtractor(FeatureExtractor fex)

Convert an edison feature extractor into an LBJ classifier

getFeatureMap

public static Map<String,Float> getFeatureMap(FeatureExtractor fex,
                                              Constituent c)
                                       throws EdisonException

Create a feature map using the specified feature extractor on the input constituent.

Throws:

EdisonException

getFeatureSet

public static Set<String> getFeatureSet(FeatureExtractor fex,
                                        Constituent c)
                                 throws EdisonException

Create a feature set using the specified feature extractor on the input constituent. Note: If there are any real valued features, they will be ignored.

Throws:

EdisonException

convert

public static Pair<int[],double[]> convert(Set<Feature> features,
                                           edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon,
                                           boolean trainingMode)

Convert a feature set into a pair of arrays of integers and doubles by looking up the feature name in the provided lexicon.

Parameters:

features - The feature set

lexicon - The lexicon

trainingMode - Should an unseen feature string be added to the lexicon? If this is false, unseen features will be given an ID whose value is one more than the number of features.

Returns:

a pair of int[] and double[], representing the feature ids and values.

getFeature

public static edu.illinois.cs.cogcomp.lbjava.classify.Feature getFeature(edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon,
                                                                         int id)

This function gets the feature from the lexicon using the id. Use this function if you use either getFeatureId or convert to avoid off-by-one errors.

getFeatureId

public static int getFeatureId(edu.illinois.cs.cogcomp.lbjava.learn.Lexicon lexicon,
                               boolean trainingMode,
                               Feature feature)

Returns the id of the feature according to the lexicon. If trainingMode is set to false and the feature is not found in the lexicon, then the function returns -1.

conjoin

public static FeatureExtractor conjoin(FeatureExtractor f1,
                                       FeatureExtractor f2)

Conjoins two feature extractors. This creates a new feature extractor that produces conjoined features.

Parameters:

f1 - One feature extractor

f2 - Another feature extractor

Returns:

A new feature extractor that produces feature conjunctions

conjunct

public static Feature conjunct(String left,
                               String right)

Conjoins two discrete features corresponding to the strings left and right.

conjoin

public static Set<Feature> conjoin(Set<Feature> feats1,
                                   Set<Feature> feats2)

Conjoins two feature sets. This produces a cross product of features. That is, every feature from the first parameter is conjoined with every feature of the second one. If either of the input sets are empty, this function returns the empty set.

Parameters:

feats1 - The first feature set

feats2 - The second feature set

Returns:

A new feature set that contains feature conjunctions

serializeFeatureSet

public static byte[] serializeFeatureSet(Set<Feature> features)
                                  throws IOException

Serializes a set of features into a byte array.

Throws:

IOException

deserializeFeatureSet

public static Set<Feature> deserializeFeatureSet(byte[] bytes)
                                          throws IOException

Deserializes a byte array into a set of features

Throws:

IOException