Appling Text Preprocessing on real time data before entering model

Hello, i’m new to knime, i built NLP Sentiment analysis model shown in the photo and i exported it as .PMML file using PMML compiler.
i would like to apply preprocessing on my real time data before feeding it to my saved model.
How can i export my preprocessing nodes or is there any other way to do it ?

Hi @Atrby,

PMML itself only supports a limited set of pre-processing steps. We do have a blog post that describes how to build a more complex PMML model with the set of operations that we currently support.

Judging from the screenshot of the workflow that you have posted, your pre-processing seems to be more involved than regular PMML transformations. One solution could be to expose a KNIME Workflow as REST endpoint using KNIME Server. In that case, you could develop a workflow that does preprocessing on input data using the Container Input (JSON) node and returning the processed data via a Container Output (JSON) node.

Best,
Stefan

2 Likes