com.aliasi.hmm

Class AbstractHmmEstimator

java.lang.Object

com.aliasi.hmm.AbstractHmm

com.aliasi.hmm.AbstractHmmEstimator

All Implemented Interfaces:

Handler, ObjectHandler<Tagging<String>>, HiddenMarkovModel, Compilable

Direct Known Subclasses:

HmmCharLmEstimator

public abstract class AbstractHmmEstimator
extends AbstractHmm
implements ObjectHandler<Tagging<String>>, Compilable

An HmmEstimator may be used to train a hidden Markov model (HMM). Training events are supplied through the ObjectHandler interface method handle(Tagging). The estimator implements an HMM, so is suitable for use in a tag-a-little, learn-a-little environment or elswhere when an adaptive HMM is required. At any point, the estimator may be compiled to an object output stream using compileTo(ObjectOutput).

Since:

LingPipe2.1

Version:

4.0.0

Author:

Bob Carpenter

Constructor Summary

Constructors

Constructor and Description
AbstractHmmEstimator(SymbolTable table) Construct an HMM estimator with the specified tag symbol table.

Method Summary

Methods

Modifier and Type	Method and Description
abstract void	compileTo(ObjectOutput objOut) Compiles a copy of this estimated HMM to the specified object output.
void	handle(Tagging<String> tagging) Train the estimator with the specified tokens, whitespaces and states.
long	numTrainingCases() Return the number of taggings handled.
long	numTrainingTokens() Returns the number of tokens handled for training.
abstract void	trainEmit(String state, CharSequence emission) Train the emission estimator with the specified training instance consisting of a state and emission.
abstract void	trainEnd(String state) Train the end state estimator with the specified end state.
abstract void	trainStart(String state) Train the start state estimator with the specified start state.
abstract void	trainTransit(String sourceState, String targetState) Trains the transition estimator from the specified transition from the specified source state to the specified target state.

Methods inherited from class com.aliasi.hmm.AbstractHmm

addState, emitLog2Prob, emitLog2Prob, emitProb, emitProb, endLog2Prob, endLog2Prob, endProb, endProb, startLog2Prob, startLog2Prob, startProb, startProb, stateSymbolTable, transitLog2Prob, transitLog2Prob, transitProb, transitProb

Methods inherited from class java.lang.Object

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

Constructor Detail

AbstractHmmEstimator

public AbstractHmmEstimator(SymbolTable table)

Construct an HMM estimator with the specified tag symbol table.

Parameters:

table - Symbol table for tags.

Method Detail

trainStart

public abstract void trainStart(String state)

Train the start state estimator with the specified start state. This increases the likelihood that the specified state will be the state of the first token.

Parameters:

state - State being trained.

trainEnd

public abstract void trainEnd(String state)

Train the end state estimator with the specified end state. This increases the likelihood that the specified state will be the state of the last token.

Parameters:

state - State being trained.

trainTransit

public abstract void trainTransit(String sourceState,
                String targetState)

Trains the transition estimator from the specified transition from the specified source state to the specified target state.

Parameters:

sourceState - State from which the transition is made.

targetState - State to which the transition is made.

trainEmit

public abstract void trainEmit(String state,
             CharSequence emission)

Train the emission estimator with the specified training instance consisting of a state and emission. This method may be used for dictionary-based training for a particular state.

Parameters:

state - State being trained.

emission - Emission from state being trained.

compileTo

public abstract void compileTo(ObjectOutput objOut)
                        throws IOException

Compiles a copy of this estimated HMM to the specified object output. Reading in the resulting bytes with an object input will produce an instance of HiddenMarkovModel, but will most likely not be an instance of the same class as the object being compiled.

Specified by:

compileTo in interface Compilable

Parameters:

objOut - Object output to which this estimator is compiled.

Throws:

IOException - If there is an I/O exception compiling this object.

numTrainingCases

public long numTrainingCases()

Return the number of taggings handled. This is simply the number of times handle(Tagging) has been called.

Returns:

The number of taggings handled for training.

numTrainingTokens

public long numTrainingTokens()

Returns the number of tokens handled for training. This is the sum of the length of token arrays in all calls to the handle(Tagging) method.

Returns:

The number of tokens handled for training.

handle

public void handle(Tagging<String> tagging)

Train the estimator with the specified tokens, whitespaces and states. The whitespaces are ignored.

For a specified tagging made up of parallel sequences of tags and tokens, this method calls:

• trainTransit(String,String) on each tag pair,
• trainEmit(String,CharSequence) on each tag/token pair,
• trainStart(String) on the first tag, and
• trainEnd(String) on the last tag.

Specified by:

handle in interface ObjectHandler<Tagging<String>>

Parameters:

tagging - Tagging from which tokens and tags are derived.