Using Intelligent Keyword Classifier questions

The Intelligent Keyword Classifier is very confusing & seems like the old chicken & egg problem – how can you have a learning data file to provide to intelligent Keyword classifier when you haven’t classified any documents yet?? I’m trying to complete the DU training final practice exercise using Community version 21.10. Even when I looked at the solution project, I have no idea how they were able to generate that intelligent classifier.json file. I reviewed other forum posts including one where someone suggested watching a video about Elastic Search. That wasn’t helpful at all. Can anyone else help? I also looked at the doc on this topic and it doesn’t explain any details that would help.

Also, when I try to “Start Training,” I get an 'invalid API key" error even though I’ve been using the same DU api key for this whole training class. Also, how can you start training when the pdf doc provided hasn’t been split yet. That happens later.

Hi @jWaltonNext2018

For the creation of the classifier.json file you will need to train your classifier. You can use the Present Valitation Station activity to manually classifiy the document by selecting the keywords you will use for classification and then include a Train Classifier Scope to save the data into the classifier.json file.


I did a video some months ago before the Form Extractor required an endpoint and a license key, but the classification and training for the keyword base classifier is still the same.
Intelligent OCR UiPath Tutorial - YouTube