I would like to bring a table into 4th normal form, using a normalization node. This would yield great speedup for any type of workflow.
After loading my data into KNIME, a lot of that data is redundant (not normalized). Mostly there are always Key-Value combinations. But also cases where several values (hierarchies) can be mapped to a key (e.g. Company->Country->Region, each also being a Key-Value combination which results in 7 values for one key which can be associated 1:1, i.e. a mapping should take O(1)). So I should be able to extract the values and only pass the keys through the KNIME flow. This would save a lot of computation time and also memory. At the end of the flow I would then add the values back to the data structure.
The only solution I see so far would be using a Joiner node. However this seems to yield very bad performance (definitely not O(1)).
Anybody able to help me with this?