NLP.Concraft.Guess

Types

data Guesser t

A guessing model.

Constructors

Guesser

Fields

schemaConf :: SchemaConf

crf :: CRF Ob t

Instances

(Ord t, Binary t) => Binary (Guesser t)

Guessing

guess :: (Word w, Ord t) => Int -> Guesser t -> Sent w t -> [[t]]

Determine the k most probable labels for each word in the sentence. TODO: Perhaps it would be better to use sets instead of lists as output?

include :: (Word w, Ord t) => [[t]] -> Sent w t -> Sent w t

Insert guessing results into the sentence.

guessSent :: (Word w, Ord t) => Int -> Guesser t -> Sent w t -> Sent w t

Combine guess with include.

Training

data TrainConf

Training configuration.

Constructors

TrainConf

Fields

schemaConfT :: SchemaConf

sgdArgsT :: SgdArgs

SGD parameters.

onDiskT :: Bool

Store SGD dataset on disk

r0T :: R0T

R0 construction method

data R0T

Method of constructing the default set of labels (R0).

Constructors

AnyInterps

See anyInterps

AnyChosen

See anyChosen

OovChosen

See oovChosen

Instances

Enum R0T

Eq R0T

Data R0T

Ord R0T

Show R0T

Typeable R0T

train

Arguments

:: (Word w, Ord t)

=> TrainConf

Training configuration

-> IO [Sent w t]

Training data

-> IO [Sent w t]

Evaluation data

-> IO (Guesser t)

Train guesser.