Modeling differences relational, semantic web
This has been an issue from the beginning of VF. VF promises to be agnostic in tech implementation, and support what people want to use, not only semantic web. REA came from a relational time, and has always basically been that kind of model. Also, most implementations so far have a relational db under the hood. UML and relational are basically compatible. VF chose to stick with the relational modeling constructs, for example resolving many-to-many relationships with a new class (associative entity), but doing this in the RDF-based ttl file that is the system of record for VF. RDF/semantic web introduces enough differences to make it annoying to use some of the relational based constructs.
This came up with Proposal, ProposedIntent, Intent - which is probably the most annoying example. And possibly the only example where we really don't need the "extra" class to contain data? (It also has a lot of flexibility, which makes it more complex than a lot of people need, those who don't want to re-use Intents on different Proposals.) A person doing a simple offers app implemented a bypass of the many-to-many, which I thought was a good experiment.
The question is: Should we explicitly include more friendly semantic web options for this kind of situation? Then the naming would be standardized. But the vocabulary would be more complex. Or possibly we could re-model that more simply, but it would be non-standard with the rest of the vocabulary, and not be relational-friendly.