Data.CRF.Chain1.Constrained

The module provides first-order, linear-chain conditional random fields (CRFs) with position-wide constraints over label values.

Data types

data Word a b

A Word is represented by a set of observations and a set of potential interpretation labels. When the set of potential labels is empty the word is considered to be unknown and the default potential set is used in its place.

Constructors

Word

Fields

obs :: Set a

The set of observations

lbs :: Set b

The set of potential interpretations.

Instances

(Eq a, Eq b) => Eq (Word a b)

(Eq (Word a b), Ord a, Ord b) => Ord (Word a b)

(Show a, Show b) => Show (Word a b)

unknown :: Word a b -> Bool

The word is considered to be unknown when the set of potential labels is empty.

type Sent a b = [Word a b]

A sentence of words.

data Dist a

A probability distribution defined over elements of type a. All elements not included in the map have probability equal to 0.

mkDist :: Ord a => [(a, Double)] -> Dist a

Construct the probability distribution.

type WordL a b = (Word a b, Dist b)

A WordL is a labeled word, i.e. a word with probability distribution defined over labels. We assume that every label from the distribution domain is a member of the set of potential labels corresponding to the word. TODO: Ensure the assumption using the smart constructor.

annotate :: Ord b => Word a b -> b -> WordL a b

Annotate the word with the label.

type SentL a b = [WordL a b]

A sentence of labeled words.

CRF

data CRF a b

A conditional random field model with additional codec used for data encoding.

Constructors

CRF

Fields

codec :: Codec a b

The codec is used to transform data into internal representation, where each observation and each label is represented by a unique integer number.

model :: Model

The actual model, which is a map from Features to potentials.

Instances

(Ord a, Ord b, Binary a, Binary b) => Binary (CRF a b)

Training

train

Arguments

:: (Ord a, Ord b)

=> SgdArgs

Args for SGD

-> IO [SentL a b]

Training data IO action

-> Maybe (IO [SentL a b])

Maybe evalation data

-> (AVec Lb -> [(Xs, Ys)] -> [Feature])

Feature selection

-> IO (CRF a b)

Resulting model

Train the CRF using the stochastic gradient descent method. The resulting model will contain features extracted with the user supplied extraction function. You can use the functions provided by the Data.CRF.Chain1.Feature.Present and Data.CRF.Chain1.Feature.Hidden modules for this purpose. When the evaluation data IO action is Just, the iterative training process will notify the user about the current accuracy on the evaluation part every full iteration over the training part. TODO: Accept custom r0 construction function.

Tagging

tag :: (Ord a, Ord b) => CRF a b -> Sent a b -> [b]

Determine the most probable label sequence within the context of the given sentence using the model provided by the CRF.

tagK :: (Ord a, Ord b) => Int -> CRF a b -> Sent a b -> [[b]]

Determine the most probable label sets of the given size (at maximum) for each position in the input sentence.

Feature selection

hiddenFeats :: AVec Lb -> [(Xs, b)] -> [Feature]

Hidden Features of all types which can be constructed on the basis of the dataset. The default set of potential interpretations is used for all unknown words.

presentFeats :: [(Xs, Ys)] -> [Feature]

Features of all kinds which occur in the dataset.