On 7/13/11 1:20 PM, [email protected] wrote:
On Tue, Jul 12, 2011 at 5:18 PM, Jörn Kottmann<[email protected]>  wrote:

Ohh, that went wrong, I posted the sketch to the wiki:
https://cwiki.apache.org/**OPENNLP/command-line-**
parameter-interfaces.html<https://cwiki.apache.org/OPENNLP/command-line-parameter-interfaces.html>


Very nice. It will help refactoring other tools.

What is the difference between BasicTrainingParams, TrainingParams and
TrainingToolParams? Are you thinking about using this interfaces as argument
for the training and evaluation API?

BasicTrainingParams is as defined in BasicTrainngParametersI.
TrainingParams is per component, e.g. like defined in 
opennlp.tools.cmdline.tokenizer.TrainingParametersI.

TrainingToolParams should contain the params which are still missing for the 
training tool,
that is the model parameter and the data parameter, since the first is not 
needed in the CVTools
and the later one has a different arg name and description.

A training tool, as opennlp.tools.cmdline.tokenizer.TokenizerTrainerTool will 
define
a new Parameters interface which extends its TrainingParams and the general 
TrainingToolParams.
The TokenizerCVTool will also define a Parameters interface which extends
BasicCrossValidatorParameters (CVParams in the sketch) and the component 
TrainingParams.

Jörn



Reply via email to