Palladian Text Classifier - Append new Model / Combine Models

Community Extensions
1

Is there anyway to combine 2 Palladian TextClassifierLearner Output into 1 Model? Like how in PMML there are ways to append 2 PMML into 1 PMML .

For example, lets say i generate a Model using PTC last month. And this month there is new data coming in with the same column name, and I would like to combine both last month’s existing model as well as a new model used in the new data to create 1 final PTC Model.

I found a TextClassifierModelToTable, but there doesnt seem to be anyway to convert back to Model format i.e its missing a TextClassifierTableToModel Node.

Is there any work around solution? Or do i need to migrate to PMML or get the PTC to relearn everything?

2

At the moment we have no node (or workarounds) for that.

I see that this would be nice to have and the simple dictionary structure of the classifier makes it relatively straightforward to implement. On the other hand, re-training a new model is quite fast.

If this is important for your usage scenario, feel free to reach out by email and we can discuss how you can support the development of such a feature: philipp@philippkatz.de

– Philipp

@Iris Could you move this to the Palladian sub forum? Thx!

3

Ahh I see, was hoping for a workaround for that.

Oh well, I guess the current solution would be to combine Excel A and Excel B into Excel C and then re-apply the textclassifierlearner node and generate a new Model.

4

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.