All tasks in Hyperscience that require human input are called Supervision or Quality Assurance.
Supervision
A submission will travel through multiple steps during the data-extraction process. At each step, if the machine's confidence falls below the thresholds specified in the flow's settings or the application settings, the system will enlist the help of a human to complete a Supervision task.
For more information on Supervision required for data extraction, see What is Supervision?
Quality Assurance
Quality Assurance (QA) is the second type of tasks where human input is required. These tasks occur after submissions are completed and therefore do not affect the speed at which a submission flows through the system. It is important to note that if an error is found during the QA process, it will not retroactively change the output of the specific field.
For more information on QA, see the article What is Quality Assurance?
Tasks
The different tasks that you may be required to do in Supervision are listed in the table below.
Task Name | Description | Type |
|---|---|---|
Document Classification Step 1 - Categorization | This task occurs when the Manual Classification Supervision flow setting is enabled, and the Structured Layout Match Threshold for an individual page is below the defined threshold value. In this task, you must specify the category value which best describes the page shown. The following options are possible:
| Supervision |
Document Classification Step 2 - Grouping documents into folders | This task is available if the Foldering feature was enabled in a previous version of Hyperscience. All documents shown can be organized into folders. Folders group together documents that should be associated with one another due to business rules or downstream requirements. | Supervision |
Document Classification Step 3 - Add Metadata Fields | This task is available if the Metadata feature was enabled in a previous version of Hyperscience. In this task, you can add metadata to folders or documents for better management of downstream processes. | Supervision |
Table Identification (ID) | This task occurs when the Table ID confidence is below the threshold specified in the Table Identification Target Accuracy flow setting. | Supervision |
Field Identification (ID) | This task occurs when the Field ID confidence is below the threshold specified in the Field Identification target accuracy flow setting, or if you have designated a field for manual identification in the Layout Editor. | Supervision |
Transcription & Consensus | When the Manual Transcription Supervision flow setting is enabled, this task can occur as a result of two different scenarios:
You should key what you observe. Validation on certain field types can be overridden, if desired. Additionally, if at least one table column and row have been identified in a Semi-structured document, the task will show a table field. You can then review or correct any values in the table that the machine has transcribed, or transcribe any values that the machine was not able to transcribe. | Supervision |
Classification QA | These QA tasks are for Semi-structured documents only. You will be asked to verify or correct the classification of a document so the system can assess its Classification accuracy. | Quality Assurance |
Transcription QA | These QA tasks are for Structured documents only. You will be instructed to transcribe a field so the system can assess accuracy. | Quality Assurance |
Field Identification QA | These QA tasks are for Semi-structured documents only. You will be asked to confirm or reject the location of a field value. | Quality Assurance |