Relational Grammar

The Relational Grammar formalism has been proposed for characterizing languages whose expressions involve more than one-dimensional arrays of characters (strings). It was motivated initially by work in understanding visual languages such as mathematics expressions and diagrams. It has since been used in other areas such as database pattern matching, improver-based design applications, and generation of multimedia documents. The basic idea is that the vocabulary of Relational Languages is composed of objects and they are grouped hierarchically into higher level expressions through relations among objects. This Relational Grammar formalism is not to be confused with the natural language formalism with the same name in which relations were presumed to be grammatical relations. There is also another formalism within the visual language community called Relation Grammars, which is distinct from the work here. In fact, there has been a long history of higher-dimensional grammar formalisms, including array grammars, tree grammars, graph grammars, and various forms of picture and shape grammars. Unfortunately, it is often difficult to compare higher-dimensional grammar formalisms because there are subtle differences in data representations and rule definitions that are often not spelled out explicitly. In the scheme of things, Relational Languages are a member of the context-free family of higher-dimensional formalisms, which makes them weaker in expressive power than most others currently being investigated.