next up previous contents
Next: Specification of Constraint-based Grammars Up: Interpretation of Constraints Previous: Interpretation of Constraints

Readable Notation

We are using a special representation for complex constraints, called matrix notation (see also [VanNoord1993]). In order to distinguish variable names from features and constants we adopt the Prolog convention that only variable names start with a capitalized letter and names of features or constants have to be written in lower case. For example the following constraints on the variable tex2html_wrap_inline11339

tabular1753

are represented in matrix notation as follows (the variables tex2html_wrap_inline10997 and tex2html_wrap_inline10999 are computed during the computation of the basic constraint):

ex1756

If variables occur only once in a matrix they are omitted. This is the case for the above example, so that it will also be written as:

ex1765

Furthermore, empty feature structures will not been shown explicitly, i.e.,

ex1772

is written as

ex1777

For feature structures encoding lists I will adopt the list notation from [Pollard and Sag1987]. The feature structure encoding of the following list

ex1783

will be written more readable using angled brackets

ex1795

The empty list then will be written as

ex1798

We will also make use of the head/tail representation of lists known from Prolog. Thus to explicitly represent the first element of a list from the rest we write

ex1800

thus, for instance, a, b, c can also be written as a|< b, c>.

The difference list of the feature structure

ex1805

will be written as

ex1820

The empty difference list will be written as

ex1824



Guenter Neumann
Mon Oct 5 14:01:36 MET DST 1998