The cost/benefit ratio of imposing semantics on data is an open area for research.
The cost of creating an index for a technical book is something O’Reilly, for example, can estimate quite closely.
What I haven’t found is a way to estimate the benefit of having such an index?
I deeply appreciate a good index but that isn’t the hard stuff that goes into a cost/benefit calculation.
Imposing semantics at a journal article level and imposing semantics on the contents of articles are two very different costs.
What measure should be used to justify either one?