So I am doing a fairly simple automation where i have a single document (India driving license) and i am using a “keyword based classifier” to classify that document.
I have provided keywords as shown below :
and also configured the classifier.
Yet i always get the below error (since classification results are empty)
I am unable to proceed beyond this.I have worked with intelligent keyword classifier.And it has worked with training.The keywords given above has been picked up from intelligent keyword classifier.
I dont see much information on keyword based classifier (for the said issue).
Can someone please help.
Hello. When selecting the document type in Validation Station you need to select some tokens from the document on the right to serve as “evidence” for that type. You can select multiple tokens by holding down the ctrl key.
keywords should NOT all be on the same line. then the document would be classified only if most of the keywords are found.
try to pun only one or a couple of keywords on a single line, as the keywords that are most often well OCRed and that define your document type.
for example try leaving only “license” as keyword.
also, try “teaching it” by creating a workflow that classifies, extracts (if necessary), shows validation station, and then trains the classifier.
it would be useful if you had a look at the Academy Document Understanding course we just launched as well, in there you will probably find more information about how classification works.
Hope this helps,
I had the same issue and solved it in two simple steps:
Select the document type category (in your case - “India License”), and then select a value from the document shown on the right. (Serves as the evidence)
After selecting the value, click on the “Kebab menu” (or the three dots, located on the right side of the document category) and select the option “Change extracted value (f+c)”
And that’s it!
Hopefully this is useful for somebody!