Generate a separation benchmark
Use the procedure in this topic to generate a separation benchmark.
If you do not have a prepared set of golden files, you can use any benchmark set to generate a separation benchmark with a little additional configuration. The document set used to perform a separation benchmark cannot contain any subfolders under the Root Folder. The folder hierarchy for a document set is visible only in the Hierarchy View.
- Open the Documents window if it is not already open.
-
In the
Documents window, on the toolbar, select
Open Document Set.
The Open Document Set window is displayed.
-
In the
Documents window, select the newly added benchmark set.
If you do not have a benchmark set, convert a custom test set into a benchmark set.
Your benchmark set expands and its documents are displayed.
-
If a different view is in use, switch to the
Hierarchy View.
The Hierarchy view is available only for test sets or benchmark sets when the default document subset is selected. If any other document subset is selected, the
Hierarchy View toolbar setting is not available.
-
In the Documents window toolbar, select
Select All.
All documents in the document set are highlighted to indicate selection.
-
On the
Process tab, in the
Test group, select
Classify.
The documents in your document set are classified.
-
On the
Documents window toolbar, select
Save All Documents.
All class assignments are saved to the document set.
-
Right-click the document set header and select
Use as Benchmark Set.
The document set is updated to a benchmark set.
-
In your document set, select the
<All Documents> document subset.
The
Hierarchy View toolbar setting is enabled.
-
On the
Process tab, in the
Benchmark group, select
Separation.
This loads the test documents into memory, merges all of the pages from all documents into a single document, and then separates this document using your existing separation settings. A progress bar indicates what percentage of the documents processed.
You can also press Ctrl+F9 to begin a separation benchmark.
When the progress bar reaches 100%, the Separation Benchmark window is displayed and includes the following results for analysis.
Any inconsistencies with the actual file structure are indicated in red text. Documents and pages with classification and separation errors are marked with an incorrect error
icon. Correct document and pages are marked with a correct
icon.
Setting
Description
Separation Result
This group displays the content structure, classification, and page count for each of the separated documents, and their desired results.
Page Separation Results
This group has the following settings:
- Page count:
-
The total number of pages in the document set.
- Correct pages:
-
The total number and percentage of correctly classified and separated pages against the total page count.
- Additional splits:
-
The total number and percentage of pages that were split unnecessarily into a new document against the total page count.
- Missed splits:
-
The total number and percentage of pages that were not split into new documents when expected against the total page count.
Document Separation and Classification Results
This group has the following document statistics:
- Document count:
-
The total number of documents in the document set.
- Correct documents:
-
The total number and percentage of correctly classified and separated documents against the total document count.
- Incorrect documents:
-
The total number and percentage of incorrectly classified and separated documents against the total document count.
- Rejected documents:
-
The total number and percentage of unsuccessfully classified, or unclassified documents against the total document count.
Analysis of Incorrect Documents
This group has the following document statistics:
- Incorrect classification:
-
The total number and percentage of incorrectly classified documents.
- Additional splits:
-
The total number and percentage of documents that were split unnecessarily against the total number of documents.
- Missed splits:
-
The total number and percentage of documents that were not split when expected against the total number of documents.