Scoring Field Output Accuracy

Field Output Accuracy is scored based on the transcription that was output.

• For Structured documents, the Field Output Accuracy is measured on the transcription output regardless of whether it is a human or machine entry.

• For Semi-structured documents, the Field Output Accuracy depends on whether the Field ID is correct. The field will first go through Field ID QA and then Transcription QA.

Scoring Field Output

If both Field ID QA and Transcription QA are enabled, then Field ID QA must be completed first before the field can be sampled for Transcription QA.

• When Field ID QA is enabled, only fields that went through Field ID QA are eligible for Transcription QA.

• Be mindful that the number of sampled fields sent to Transcription QA depends on the sample rate set for Field ID QA.

• For example, if the sample rate for Field ID QA is set to 50% and the Transcription QA sample rate is set to 100%, the sampled fields will be 100% of Field ID QA tasks, but not 100% of Field ID Supervision tasks.

• If the field is sampled for Transcription QA and the output location is found to be wrong during Field ID QA, no additional Transcription QA tasks are created for that field. The output location is automatically scored as incorrect, and transcription accuracy will not be scored.

If Field ID QA is disabled, fields on Semi-structured documents can still be sampled for Transcription QA. Field Output Accuracy can be measured but will not be accurate.

If Field ID QA is enabled and Transcription QA is disabled, then system output accuracy cannot be measured.