6.4.2 The combine-rules
Now let's have a look at the
combine-rules that pllel the syntax rules and combine the constraint graphs for subphrases to the constraint graph for a larger phrase.
Now let's have a look at the combine-rules that parallel the syntax rules and combine the constraint graphs for subphrases to the constraint graph for a larger phrase.
The First Node is the Root
The general principle is that each constituent of the sentence is associated with a part of the final constraint graph. We're going to maintain the invariant that the first node in the node list of such a partial graph is the root of this subgraph. The graphs for different constituents are combined by adding constraints that relate their roots.
One simple but central predicate we make use of here is mergeUSR/2 (see cllsLib.pl). It combines constraint graphs by merging their respective lists; the root of the merged graph will be the root of the first (leftmost) graph that was given to mergeUSR/2. Best you do Exercise 6.4 right away in order to become familiar with this predicate.
Most of the combine-rules are trivial. They simply lift the semantics of some syntactic category to that of a higher category without adding any further material. As an example, see the rule 
:
combine(np1:A,[pn:A]).
We present the more complex combine-rules below, except for one (namely 
). We leave the formulation of this rule to you as Exercise 6.5.
, 
, and 