org.imixs.ml.service

Class TrainingService

java.lang.Object

org.imixs.ml.service.TrainingService

public class TrainingService
extends Object

The TrainingService loads a list of documents and creates a TraingData object for each document.

The TrainingService extracts the text contained in a document attached to a snapshot workitem. The file content is created by a OCR scan.

If a valuable training data set can be build for a document, than the XMLTraingData object is send to the Imixs-ML service to train a ml-model.

Version:

2.0

Author:

rsoika

Field Summary

Fields

Modifier and Type	Field and Description
protected javax.enterprise.event.Event<EntityObjectEvent>	entityObjectEvents
static String	FILE_ATTRIBUTE_TEXT
static String	ITEM_ML_DEFINITIONS
static String	ITEM_ML_ITEMS

Constructor Summary

Constructors

Constructor and Description
TrainingService()

Method Summary

Modifier and Type	Method and Description
List<org.imixs.workflow.ItemCollection>	getMLDefinitions(org.imixs.workflow.ItemCollection workitem) This method returns a list of ItemCollection object holding the ml.definitions stored in a given workitem.
void	printXML(XMLTrainingData trainingData) Log the training data into the server log
void	testWorkitemData(org.imixs.workflow.ItemCollection config, org.imixs.workflow.ItemCollection doc, org.imixs.melman.WorkflowClient workflowClient) This method is used to test an existing model.
MLTrainingResult	trainWorkitemData(org.imixs.workflow.ItemCollection config, org.imixs.workflow.ItemCollection workitem, org.imixs.melman.WorkflowClient workflowClient) This method is used to extract the text contained in a snapshot document and search for 'known' entities within the text.

Methods inherited from class java.lang.Object

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

Field Detail

FILE_ATTRIBUTE_TEXT

public static final String FILE_ATTRIBUTE_TEXT

See Also:

Constant Field Values

ITEM_ML_ITEMS

public static final String ITEM_ML_ITEMS

See Also:

Constant Field Values

ITEM_ML_DEFINITIONS

public static final String ITEM_ML_DEFINITIONS

See Also:

Constant Field Values

entityObjectEvents

@Inject
protected javax.enterprise.event.Event<EntityObjectEvent> entityObjectEvents

Constructor Detail

TrainingService

public TrainingService()

Method Detail

trainWorkitemData

public MLTrainingResult trainWorkitemData(org.imixs.workflow.ItemCollection config,
                                          org.imixs.workflow.ItemCollection workitem,
                                          org.imixs.melman.WorkflowClient workflowClient)

This method is used to extract the text contained in a snapshot document and search for 'known' entities within the text. If a valuable training data set can be build the training data is send to the Imixs-ML service.

The method generates statistical data.

Parameters:

config - - a config object providing the training configuration

workitem - - a workitem providing the data

workflowClient - - a rest client instance

Returns:

- quality result

testWorkitemData

public void testWorkitemData(org.imixs.workflow.ItemCollection config,
                             org.imixs.workflow.ItemCollection doc,
                             org.imixs.melman.WorkflowClient workflowClient)

This method is used to test an existing model. The method extracts the text contained in a snapshot document and sends the text to the Imixs-ML service to be analyzed. The resuls are printed out.

Parameters:

doc - - a workitem providing the attachments and the entity data

items - - String list with items

workflowClient - - a rest client instance

printXML

public void printXML(XMLTrainingData trainingData)

Log the training data into the server log

Parameters:

trainingData -

getMLDefinitions

public List<org.imixs.workflow.ItemCollection> getMLDefinitions(org.imixs.workflow.ItemCollection workitem)

This method returns a list of ItemCollection object holding the ml.definitions stored in a given workitem. Each endpoint definition is defined by a set of items stored in a map. The Map can be conferted into a ItemCollection