de.fu_berlin.ties.extract.amend
Class TrainingContextDetails

java.lang.Object
  extended by de.fu_berlin.ties.text.TokenDetails
      extended by de.fu_berlin.ties.extract.amend.ContextDetails
          extended by de.fu_berlin.ties.extract.amend.TrainingContextDetails

public class TrainingContextDetails
extends ContextDetails

Extends the ContextDetails class by also storing the expected (true) and predicted state on of a token. Instances of this class are immutable and thus thread-safe.

Version:
$Revision: 1.1 $, $Date: 2004/11/19 14:04:51 $, $Author: siefkes $
Author:
Christian Siefkes

Constructor Summary
TrainingContextDetails(String theToken, int tokenRep, int tokenIndex, boolean wsBefore, FeatureVector features, CombinationState predicted, CombinationState expected)
          Creates a new instance.
TrainingContextDetails(TokenDetails orgDetails, FeatureVector features, CombinationState predicted, CombinationState expected)
          Creates a new instance, re-using field values from a TokenDetails instance.
 
Method Summary
 CombinationState getExpectedState()
          Returns the expected (true) CombinationState for this instance.
 CombinationState getPredictedState()
          Returns the CombinationState predicted by the current classification model for this token if there was an classification error; or null if the expected state was correctly predicted.
 
Methods inherited from class de.fu_berlin.ties.extract.amend.ContextDetails
getContext
 
Methods inherited from class de.fu_berlin.ties.text.TokenDetails
getIndex, getRep, getToken, isWhitespaceBefore, toString
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Constructor Detail

TrainingContextDetails

public TrainingContextDetails(String theToken,
                              int tokenRep,
                              int tokenIndex,
                              boolean wsBefore,
                              FeatureVector features,
                              CombinationState predicted,
                              CombinationState expected)
Creates a new instance.

Parameters:
theToken - the token to represent
tokenRep - the repetition of the token in the document (counting starts with 0, as the first occurrence is the "0th repetition")
tokenIndex - the index of the token in the document (counting starts with 0 for the very first token)
wsBefore - whether there is whitespace before the main token (either at the end of left or in the preceding element)
features - a feature vector representing the context of the token
predicted - the CombinationState predicted by the current classification model for this token if there was an classification error; or null if the expected state was correctly predicted
expected - the expected (true) CombinationState for this token

TrainingContextDetails

public TrainingContextDetails(TokenDetails orgDetails,
                              FeatureVector features,
                              CombinationState predicted,
                              CombinationState expected)
Creates a new instance, re-using field values from a TokenDetails instance.

Parameters:
orgDetails - an instance of the "grandparent" class whose member fields are re-used to initialize this instance
features - a feature vector representing the context of the token
predicted - the CombinationState predicted by the current classification model for this token if there was an classification error; or null if the expected state was correctly predicted
expected - the expected (true) CombinationState for this token
Method Detail

getExpectedState

public CombinationState getExpectedState()
Returns the expected (true) CombinationState for this instance.

Returns:
the value of the attribute

getPredictedState

public CombinationState getPredictedState()
Returns the CombinationState predicted by the current classification model for this token if there was an classification error; or null if the expected state was correctly predicted.

Returns:
the value of the attribute


Copyright © 2003-2004 Christian Siefkes. All Rights Reserved.