Skip to content

Categorization

The Categorization tool window shows the categorization results of the last analysis carried out on the test document specified in the context bar.

win

Its panels are:

  • Categories
  • Hits/Blocks
  • Rule preview

A context menu command is available on the file name:

Name Description
Copy File name Copy the file name
Copy File full path Copy the file full path

Categories panel

This panel shows the categories that were returned by the categorization process and—as a detail for each category—the rules, in the form of snippets, that determined that outcome.

There's a row for each output category.
The W symbol marks "winners", the L symbol marks "losers". "Losing" categories are those that, through scripting, were ruled out of final output, but can still be visualized in Studio.

The columns are:

Column Description
Category/Rules Category name or rule's condition
Label (Only for categories) Category description
Score The overall score of the category or, for rules' snippets, the contribution of the rule to the score
Frequency (Only for categories) The score computed as a percentage of the sum of the scores of all the categories
Hits (Only for rules) The number of times the rule was matched
Compound (Only for categories) Compound score
Quality (Only for categories) Quality check result

The quality check result appears only if the category has been set as a categorization target for the test document through annotation.

The Compound column appears only if in the Studio Settings window, the Tool Windows > Categorization > Show Compound column in categories tree configuration property is set to true. For a description of the compound score read about the CHIILD_TO_FATHER option in the Studio languages reference manual.

These are the icons used to describe quality check results:

Icon Name Description
Green hexagon TP True positive, the annotated target was matched
Orange hexagon FP False positive, unexpected result, no annotation for this result
Purple hexagon FN False negative, there were no results for the annotated target
Quality undefined Incomplete match between actual extractions and targets: some values match annotations, other values don't

The toolbar contains:

Icon Name Description
Expand All Expand all the categories, showing rules
Collapse All Collapse all the categories, hiding rules
Search box Search criteria
Reset Reset the search criteria
Find Next Display the results matching the search criteria in the Find Results tool windoow
Toggle Label Visibility Display or hide the Label column
Toggle D Hit Categories Visibility Display or hide losing categories
Toggle Annotations Visibility Display or hide the annotation toolbar

The counter at the bottom of the panel shows the number of categories.

Context menu commands are:

Command Description
Copy Category Name Copy the category name
Copy Category Label Copy the category label
Copy Category Frequency Copy the category frequency
Copy Category Quality Copy the category quality
Copy Selected Row Copy the selected row
View in Taxonomy Open the Classes tool window, Taxonomy panel, and set its focus to the selected category
Copy All Selected Rows Copy all selected rows
Copy All Rows Copy all rows

Available mouse commands are:

Command Description
Click a column header except Quality Change sort order
Click a category Highlight all rule's hits in the test document inside the editor
Click a rule's snippet Highlight the rule's hits in the test document inside the editor and update the other panels
Double-click a rule snippet Like single click plus a toggle action if the command is repeated: highlight the rule in the rules file inside the editor or highlight the rule's hits in the test document inside the editor

Annotation toolbar

This toolbar allows annotating results as categorization targets for future quality checks.

Name Description
Annotate box Name of the category to annotate. The box is filled automatically with the name of the selected output category, but the value can also be set by hand.
Annotate Annotate the specified category
Annotate All Annotate all the output categories as target results

Hits/Blocks panel

This panel shows all the hits of the rule that's selected in the categories panel. For every hit, it can show the blocks, that is, the rule's sub-conditions that were satisfied and thus determined the hit.

There's a row for each hit.
Score is the contribution the hit gave to the overall category score.

The toolbar contains:

Icon Name Description
Expand All Expand all the hits, showing blocks
Collapse All Collapse all the hits, hiding blocks

The counter at the bottom of the panel shows the number of hits for the selected rule.

Available mouse commands are:

Command Description
Click a hit Highlight the hit in the test document inside the editor
Click a block Highlight the text that was matched by the block in the test document inside the editor
Double-click a hit Toggle action if the command is repeated: highlight the rule in the rules source file inside the editor or highlight the rule's hits in the test document inside the editor

Rule preview panel

This panel shows the rule that's selected in the categories panel, without the need to open the rule's source file in the editor.