Question
Areteans
AU
Last activity: 9 May 2019 11:10 EDT
How to train the NLP models to learn automatically from the feedback provided?
Hi Support Team,
I have read the below articles and aware of how the models can be updated with feedback loop manually by a data scientist post reviewing the data.
https://www1.pega.com/products/pega-platform/intelligent-virtual-assistant/email
However, can anyone let me know if there is a way to update the models with the feedback data automatically without having the intervention of any user?
Below is the scenario where I am currently working on :
Run the NLP models(Topic detection) on the input data and try to match relavant category based on keywords provided in the Taxonomy sheets. If the category is not found the case routes to a user who will update the relavant category and move the case ahead. This data will be written into the Training models as well.
The only way I see to put this feedback data into the model to learn is to run the update model wizard again manually. Is there an automated way to input the feedback data into model without having to run the update model wizard manually?
PEG
GB
I haven't enough exposure to our NLP feature to say if there is an OOTB way.
Having said that, the wizard you are talking about is certainly based on a Pega process . So in therory you could copy this wizard and replace manual steps (assignment) for automatic steps (utilities)
Pegasystems Inc.
IN
In 74, you'll need to run Update model wizard from Analytics center. In Infinity, we will have the facility to fire build from within Training data tab.
Wipro Ltd
US
Do we have any examples /working code that updates NLP model / Text categorization / Topic ?
In my activity I am calling pxCaptureTAFeedback followed by pxUpdateModels. However I do not see expected results.
Thanks
Al
Pegasystems Inc.
IN