given a model of a theory over a 1st-order language, the set of atomic sententences and negations thereof of that hold in the model, when the language is extended with constant symbols for each element in the domain in the model
Discovered by embedding cosine similarity (sentence-transformers MiniLM, 384-dim).