Declarative specification of visual languages

In order to specify the relationship between pictures and their meaning, a class of declarative, constraint-based picture specification languages is introduced. These extend previous approaches to picture specification based on constraints and grammar formalisms. As the languages are defined as an abstract language scheme, it is ensured that they share a common declarative semantics and execution model. The declarative semantics allows both people and machines to easily reason about a specification independently of any underlying implementation. The operational semantics permit a picture specification to be used to both generate and recognize pictures defined by visual languages. Implementation techniques developed for constraint logic programs, which enable this to be done efficiently, are discussed.<>