Hello all, we’re building a Comms Minning project, but we cannot get any prediction from signature that is automatically generated by the tool from email’s body. We use general fields extraction. Is there any way either avoid the split or generate prediction on signature? thanks Pawel
Hey @Szlamka_Pawel ,
You’re absolutely right to raise this – it’s a tricky one. Communications Mining does automatically split out the email signature and excludes it from both classification and prediction by design.
At the moment, there’s no documented way to disable this behavior or to force predictions on the signature block.
One workaround would be to preprocess the email content yourself - merging the signature back into the main body before it’s sent into Communications Mining. You could also consider handling it through a separate extraction flow if needed.
Alternatively, it may be worth submitting this as a feature request to UiPath or reaching out through your account manager.
Hey @pikorpa ,
I was afraid that it would not be possible. I wonder about the workaround that you suggested, but I upload the whole email body to the Comms Minning and then the split is done by the tool itself, so how this preprocessing should be done?
thanks
Pawel
Hey @Szlamka_Pawel ,
You’re absolutely right - after digging deeper into how Communnication Mining handles internal pre-processing, I have to admit this isn’t as straightforward as I initially thought. Even if we merge the signature into the body before submission, the tool still runs its own signature detection algorithm and may split it off anyway - meaning it gets excluded from prediction either way.
So yes, at this point there doesn’t seem to be a reliable way to force the model to include the signature content, unless UiPath provides an option to disable it. I’d suggest raising a request to UiPath support to see if there’s away to control or disable the signature detection behavior.
This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.