The idea of a "Semantics Extensibility Point" is that it's reasonable for the a syntactic expression to have multiple distinct meanings and that one should indicate the intended meaning.
Notes on this from the F2F1 Breakout Session on OWL and RDF Compatibility:
- This might take the form of a "which semantics are you using" flag at the top of a RIF document, or be attached to individual syntactic constructs (there could be multiple "implies" operators, with different semantics.)
During Phase 1, we should only need two semantics: FOL Semantics and Minimal-Model Semantics. The group agreed that differences between those semantics would manifest in response to SPARQL queries, but that differences among other possible semantics would not manifest in Phase 1.
- The relationships between all standard semantics need to be spelled out clearly, so people know if they use the "wrong" semantics whether they'll get incomplete reasoning, unsound reasoning, etc. (EG a sound and complete MMS reasoner will be sound but incomplete on an FOL ruleset.)