Package nltk :: Package classify :: Module maxent :: Class GISEncoding

type GISEncoding

             object --+        
                      |        
 MaxentFeatureEncodingI --+    
                          |    
BinaryMaxentFeatureEncoding --+
                              |
                             GISEncoding

A binary feature encoding which adds one new joint-feature to the joint-features defined by BinaryMaxentFeatureEncoding: a correction feature, whose value is chosen to ensure that the sparse vector always sums to a constant non-negative number. This new feature is used to ensure two preconditions for the GIS training algorithm:

At least one feature vector index must be nonzero for every token.
The feature vector must sum to a constant non-negative number for every token.

Instance Methods

[hide private]

__init__(self, labels, mapping, unseen_features=False, alwayson_features=False, C=None)

source code

list of (int, number)

encode(self, featureset, label)
Given a (featureset, label) pair, return the corresponding vector of joint-feature values.

source code

int

length(self)
Returns: The size of the fixed-length joint-feature vectors that are generated by this encoding.

source code

str

describe(self, f_id)
Returns: A string describing the value of the joint-feature whose index in the generated feature vectors is fid. source code

Inherited from BinaryMaxentFeatureEncoding: labels

Class Methods

[hide private]

Inherited from BinaryMaxentFeatureEncoding: train

Instance Variables

[hide private]

_C
The non-negative constant that all encoded feature vectors will sum to.

Inherited from BinaryMaxentFeatureEncoding (private): _alwayson, _labels, _length, _mapping, _unseen

Properties

[hide private]

C
The non-negative constant that all encoded feature vectors will sum to.

Method Details

[hide private]

init(self, labels, mapping, unseen_features=False, alwayson_features=False, C=None)
(Constructor)

source code

Parameters:

C - The correction constant. The value of the correction feature is based on this value. In particular, its value is C - sum([v for (f,v) in encoding]).

Overrides: BinaryMaxentFeatureEncoding.__init__

encode(self, featureset, label)

source code

Given a (featureset, label) pair, return the corresponding vector of joint-feature values. This vector is represented as a list of (index, value) tuples, specifying the value of each non-zero joint-feature.

Returns: list of (int, number)
Overrides: MaxentFeatureEncodingI.encode: (inherited documentation)

length(self)

source code

Returns: int: The size of the fixed-length joint-feature vectors that are generated by this encoding.
Overrides: MaxentFeatureEncodingI.length: (inherited documentation)

describe(self, f_id)

source code

Returns: str: A string describing the value of the joint-feature whose index in the generated feature vectors is fid.
Overrides: MaxentFeatureEncodingI.describe: (inherited documentation)

Property Details

[hide private]

C

The non-negative constant that all encoded feature vectors will sum to.

type GISEncoding

__init__(self, labels, mapping, unseen_features=False, alwayson_features=False, C=None) (Constructor)

encode(self, featureset, label)

length(self)

describe(self, f_id)

C

init(self, labels, mapping, unseen_features=False, alwayson_features=False, C=None)
(Constructor)