I don't think I've seen any reference to this in certain circles but have seen other DV books talk about this and am wondering if it's just a difference in what it's called or what we mean by 'unit of work' or a more fundamental difference of opinion.
I believe the example given was a link describing a sale in a store. The link was between customer, store, and sales representative but NOT products purchased since that was of different cardinality and thus a separate link. So if customer A, store B, and sales rep C were involved in a sale X purchasing items 1 and 2, in the relative cardinality world we would have one link table relating X-A-B-C and another link table relating X-1 and X-2. If we ignore relative cardinality I think we'd simply have one link table with two entries: X-A-B-C-1 and X-A-B-C-2.
The first seems more elegant when many products could be involved, but in other examples that are not sales related it seems like another table and join for not much benefit other than reducing redundancy/ storage, which is no longer a high priority.
Would like to hear people's thoughts on this.