7.00.02 - NERTrainer Arguments - Aster Analytics

Teradata Aster® Analytics Foundation User GuideUpdate 2

Product
Aster Analytics
Release Number
7.00.02
Release Date
September 2017
Content Type
Programming Reference
User Guide
Publication ID
B700-1022-700K
Language
English (United States)
TextColumn
Specifies the name of the input table column that contains the text to analyze.
ExtractorJAR
[Optional] Specifies the name of the JAR file that contains the Java classes that extract features. You must install this JAR file in Aster Database under the user search path before calling the function. The function includes the predefined extractor classes described in the table in Feature Template. Default behavior: The function uses only the predefined extractor classes.
The name jar_file is case-sensitive.
FeatureTemplate
Specifies the name of the file that specifies how to generate features when training the model. You must install this feature template file in Aster Database under the user search path before calling the function. For more information about template_file, see Feature Template.
ModelFile
Specifies the name of the model file that the function generates and installs in Aster Database.
Language
[Optional] Specifies the language of the input text:
  • 'en': English (Default)
  • 'zh_CN': Simplified Chinese
  • 'zh_TW': Traditional Chinese
MaxIterNum
[Optional] Specifies the maximum number of iterations. Default: 1000.
Eta
[Optional] Specifies the tolerance of the termination criterion. Defines the differences of the values of the loss function between two sequential epochs. Default: 0.0001.

When training a model, the function performs n-times iterations. At the end of each epoch, the function calculates the loss or cost function on the training samples. If the loss function value change is very small between two sequential epochs, the function considers the training process to have converged.

The function defines Eta as:

Eta=(f(n)-f(n-1))/f(n-1)

where f(n) is the loss function value of the nth epoch.

MinOccurNum
[Optional] Specifies the minimum number times that a feature must occur in the input text before the function uses the feature to construct the model. Default: 0.