I’m using KNIME to merge 2 sets of complex objects that represent chemical substances (names, identifiers, structure, etc.).
Currently, my workflow reads in 2 sets of JSON strings, converts to a JSON column, uses the versatile JSON Path node to create strings, then compares/evaluates/manipulates sets of strings to merge the data into one set and converts them back to JSON. It’s a complex process and prone to error.
I’m imagining something that would make the process much easier and less prone to error:
Being able to define custom objects that could be deserialized from JSON, manipulated using KNIME nodes that filter, join, merge, transform, etc. and then turned back into JSON at the end of the process.
Has anyone thought about providing the ability to define one’s own (complex) objects in KNIME in order to do this kind of processing?