The Suspect Level is the level of suspicion placed in the Keyword Values captured during Intelligent Capture for AP processing (ICAP processing). Suspect Keyword Values must be manually reviewed in index verification.
When a document is processed, the OCR engine gives each extracted Keyword Value a score between 0 and 100, depending on how suspicious the engine is of the OCR quality of the returned result. Lower scores indicate that the OCR engine is less suspicious of the extracted values, whereas higher scores indicate that the OCR engine is more suspicious of the extracted values.
The Suspect Level threshold is a configurable setting that specifies the highest limit at which a returned value's Suspect Level score is considered acceptable by the OCR engine. Any value with a score below the Suspect Level threshold is marked as acceptable, and any value with a score that exceeds the Suspect Level threshold is marked as suspect. For example:
-
Setting the Suspect Level threshold to 100 indicates that all results will be marked as acceptable, because all results would fall below the threshold limit. No manual verification would be required since all results are considered acceptable.
-
Setting the Suspect Level threshold to 0 indicates that all results will be marked as suspect, because all results would exceed the threshold limit. Manual verification would be required for every value since all results are considered suspect.
-
By default, the Suspect Level threshold is set to 70. Any captured value with a Suspect Level score of 70 or lower would be marked as acceptable, and any value with a score that exceeds 70 would be marked as suspect. Manual verification would only be required for values whose scores exceed the Suspect Level threshold of 70.
The Suspect Level threshold can be configured by an administrator in the Hyland.DataCapture.Server.exe.config file. For more information about this file and about configuring the Intelligent Capture for AP engine's Suspect Level threshold, contact your system administrator.