As it relates to content classes, I am curious about the performance statistics of content class extraction stages which contain two (2) content classes vs. the performance statistics of two (2) content class extraction stages, each containing a single content class.
Use-Case: I recently posted a question around the idea of linking multiple pipelines: Multiple Pipelines in Workflows
I am curious now, if it would be to my advantage to extract all of the common functions (like content class extraction) to a common precursor pipeline; or, if by doing so, I would be adding unnecessary processing time to the workflow. In my specific use-case, I have a considerable amount of precursor processing (extraction, detection, tokenization, mapping) that can be reduced to a single pipeline (rather than be repeated across all of my pipelines.)
Anyone have experience in this area?