Use feedback to retrain a classification model.
Before you begin
The administrator must create a classification model with IBM®
StoredIQ® Administrator. For information about creating a model, see Building an auto-classification model.
To be able to select categories for a data object, you must have the standard Data User role or any other role that permits previewing data objects.
Procedure
- From IBM
StoredIQ Data Workbench,
create a user infoset.
-
Select the user infoset and run the Step-up Snippet in the Action
pane.
-
Select an enhancement in the Enhance pane to apply to the user infoset
and then click Run Enhancement.
- Create a filter and apply the filter:
- Select the Auto-classify option.
- In the list below the Auto-classify option,
select the classification model for which results are filtered.
- Select the categories. By default, the Category option
is selected, as are all of the available categories.
- Determine how the results are displayed, selecting either And
the highest score in the selected category or And
where the score <is/Greater than/Less than/Not equal/Greater than
or equal/Less than or equal> <1>.
- Click Preview Results.
- Select one of the returned objects to view in the Data
Object Viewer.
- Click the button to the right of Auto-class
Scores.
A list of categories and score
of the object for each category appear. The category with the highest
score is listed at the top.
- Select the categories to which the selected data object
belongs.
Data objects with a value closer to one (1) are
more closely associated with that category. Data objects with a value
closer to zero (0) are not associated with that category.
- Click Submit.
This feedback
helps improve the accuracy and validity of the classification model.
Repeat this step for as many different data objects as you like. The
number of feedback submissions can be seen within the <model
name> Details panel in IBM
StoredIQ Administrator.
-
At this step, the administrator needs to select the classification model from IBM
StoredIQ Administrator and start the retraining process for this model.
If the model was uploaded without a learning archive (SARC file) or if it received no
feedback, then the Retrain button is disabled.
- From IBM
StoredIQ Data Workbench,
run the enhancement against the user infoset again to see the new
scores within the Data Object Viewer.
The
improved scores indicate greater validity and accuracy.