"context" is not yet explicitely modelled, therefore we use concepts in here. The concept types are not yet enforced either. Maybe using additional tools like XSLT (Schematron?) to check this constraint is easier. No keyrings are yet set up, XML Schema allows to define distinct sets of identifiers. Not yet modelled (do we want to?) Not yet modelled (do we want to?) Not yet modelled (do we want to?) Type is modelled as attribute, the notion of an explicit "referent" has been dropped. quantifier could be modelled as attribute, defaulting to "existential". We would need some way to represent the defined quantifiers then and it will be harder to validate. This is different compared to section 6.7 of the standard draft, where the universal quantifier is a defined quantifier. How to model this one? (TODO) An empty element represents an undetermined designator. We still don't understand the distinction between literals and locators (TODO) "graph" is modelled as element so it can be used as starting element The content of this has to match an id of a concept this could be dropped in favour of forcing a declaration as separate construct somewhere The content of this has to match the id of a parameter in a lambda expression If at least one parameter is given the graph is a lambda expression (ok? what about 0-adic lambda expressions?). The type is modelled as attribute, the number and order is defined by the XML structure. has type (required) and id (optional) as attribute. The id can be used for coreferences, maybe it should be required.