Inference and Variables:

A system must be able to use inference—to draw valid conclusions based on the meaning representation of inputs and its background knowledge. Now consider the following request: I’d like to find a restaurant where I can get vegetarian food. Answering this request requires the use of variables, using some representation like the following: Serves(x,VegetarianFood). Matching succeeds only if the variable x can be replaced by some object in the knowledge base in such a way that the entire proposition will then match.

Expressiveness:

A meaning representation scheme must be expressive enough to handle a wide range of subject matter, ideally any sensible natural language utterance.