Static Private Member Functions | |
_.Library.Status | RunModelFromDomain (_.Library.Integer pResultsId, _.Library.String pDefinitionClass, _.Library.Integer pDomainId, _.Library.String pMetadataField, _.iKnow.Filters.Filter pFilter, _.Library.String pResultsClass, _.Library.Boolean pTracking, _.Library.String pKeepScoreForClass) |
_.Library.Status | RunModelFromResultSet (_.Library.Integer pResultsId, _.Library.String pDefinitionClass, _.Library.ResultSet pResultSet, pFieldMap, _.Library.String pResultsClass, _.Library.Boolean pTracking) |
_.Library.Status | RunModelFromSQL (_.Library.Integer pResultsId, _.Library.String pDefinitionClass, _.Library.String pSQL, pFieldMap, pSQLParams, _.Library.String pResultsClass, _.Library.Boolean pTracking) |
The InterSystems IRIS NLP iKnow technology is now deprecated. Please see the product documentation for more detail.
This class bundles a number of methods for validating Text Classifiers against existing data.
|
staticprivate |
Validates a Text Classifier against all sources in a domain, optionally filtered through a
<class>iKnow.Filters.Filter</class> filter object pFilter, checking the model outcome against the actual value found in the metadata field named pMetadataField.
Use pResultsClass to define the class (subclass of <class>DeepSee.PMML.Utils.Result</class>) the results of this batch run will be written to. This defaults to <class>DeepSee.PMML.Utils.TempResult</class>, which stores its data in the Temp database and therefore gets purged upon restarting the instance.
Results will be stored with their TestId property set to the value of pResultsId if a nonzero, non-null value is provided, or call the GetNewTestId() method of pResultsClass to generate a new one.
If pTracking is true, progress information will be written to ^IRIS.Temp.PMML.BgrTask($job), so this method can be run in the background and tracked by a UI.
|
staticprivate |
Validates a Text Classifier against all records returned by a custom ResultSet, looping through
the rows and checking the model outcome against the actual value found in the query result. The ResultSet should either return columns named _Text and _Category containing the text to categorize and the actual category value, or a mapping from these mandatory fields to the column names should be provided through pFieldMap.
Use pResultsClass to define the class (subclass of <class>DeepSee.PMML.Utils.Result</class>) the results of this batch run will be written to. This defaults to <class>DeepSee.PMML.Utils.TempResult</class>, which stores its data in a temporary global and therefore gets purged upon restarting the instance.
Results will be stored with their TestId property set to the value of pResultsId if a nonzero, non-null value is provided, or call the GetNewTestId() method of pResultsClass to generate a new one.
If pTracking is true, progress information will be written to ^IRIS.Temp.PMML.BgrTask($job), so this method can be run in the background and tracked by a UI.
|
staticprivate |
Validates a Text Classifier against all records returned by a user-supplied SQL query, looping through
the rows and checking the model outcome against the actual value found in the query result. The SQL query should either return columns named _Text and _Category containing the text to categorize and the actual category value, or a mapping from these mandatory fields to the column names should be provided through pFieldMap.
Use pResultsClass to define the class (subclass of <class>DeepSee.PMML.Utils.Result</class>) the results of this batch run will be written to. This defaults to <class>DeepSee.PMML.Utils.TempResult</class>, which stores its data in the Temp databsae and therefore gets purged upon restarting the instance.
Results will be stored with their TestId property set to the value of pResultsId if a nonzero, non-null value is provided, or call the GetNewTestId() method of pResultsClass to generate a new one.
If pTracking is true, progress information will be written to ^IRIS.Temp.PMML.BgrTask($job), so this method can be run in the background and tracked by a UI.