I am currently using the Document Understanding tool to extract information from documents by using form extractor. However, in the validation station, I encounter the message ‘Please select evidence for the field,’ and data in other fields is not being extracted at all. What could be the reason for this?
The message “Please select evidence for the field” in the validation station of UiPath Document Understanding typically indicates that the field requires additional validation or confirmation before the extraction process can proceed. This message appears when the extraction confidence level is below a certain threshold or when there is ambiguity in the extracted data that needs human validation.
Here are some possible reasons and solutions for encountering this message:
Low Confidence Level: The extraction confidence level for the field might be below the configured threshold. You can adjust the confidence threshold in the Form Extractor configuration to control when the validation station should prompt for human validation. Increasing the confidence threshold may reduce the frequency of this message, but it could also result in fewer fields being extracted automatically.
Ambiguous Data: The extracted data may be ambiguous or unclear, leading to the need for human validation. Ensure that the data in the document is consistent and formatted properly to improve extraction accuracy. You can also configure pre-processing steps, such as OCR preprocessing or image enhancement, to improve the quality of the extracted data.
Missing Data or Configuration: Double-check the configuration of your form extractor to ensure that all required fields are properly configured and mapped. Missing or incorrectly configured fields can lead to incomplete extraction or validation issues.
Complex Document Structure: If the document has a complex structure or layout, it may require additional processing or configuration in the form extractor to accurately extract the data. Consider adjusting the extraction settings or implementing custom extraction rules to handle complex document layouts.
Human Validation: In some cases, manual validation and correction of extracted data may be necessary to ensure accuracy. Train your validation station operators to review and validate the extracted data, providing evidence or correcting errors as needed.
By addressing these potential issues and fine-tuning your Document Understanding workflow, you can improve the accuracy and efficiency of data extraction and validation in UiPath.
I think you still have to define classification for this document type (i.e Keywords to define this document) . I would suggest you to try with different OCR’s in Digitize document stage.
