engleski

Towards a Framework for Specifying Visual Languages

Visual programming has become an important area of research in recent years, motivated by the belief that pictures are as a means of communication more powerful than words. The general notion of the visual programming paradigm is that an arrangement of graphical expressions in two, or possibly more, dimensions represents a program. The strength of visual languages lies in their ability to enable programming with graphical expressions, providing thus easier and more natural communication between man and computer. As the field is rapidly maturing, there is by now a number of different programming systems developed that employ the visual paradigm, thus influencing the refinement of adequate definition techniques. Most approaches to visual language specification that have been so far proposed in the literature can be assigned to one of the following three categories: grammatical, logical and algebraic formalisms. It is already noted that no unified approach to visual language theory has been found. A suitable framework identifying key features influencing specification expressiveness could thus serve to unite the different approaches in visual programming theory research, as well as to facilitate the use of theoretical results in visual languages specification. As a first step in that respect a framework for the most commonly used formalism – the grammar-based one – is presented in the paper. Those features manifest in the most referenced grammar-based visual language specifications known from the open literature are included in this framework: underlying conceptual model, set of graphical (primitive) objects, respective attributes and constraints in composing complex graphical objects, spatial relations among them, and the application domain. The referent grammar-based specification techniques used for assessing the proposed framework include the following ones: picture specification language, picture layout grammars, attribute icon-replacement grammars, picture processing grammars, relation grammars, and extended unification-based grammars. The picture specification language is based on a declarative, constraint-based approach defining classes of pictures, each of which consists of certain graphical objects having some structural relationships and satisfying some constraints. The picture layout grammar is based on a new model of grammar, which generates sets of objects with attributes. Within this approach a visual sentence is an unordered collection – a multiset – of visual symbols with attributes containing positional information. The conceptual model of the attribute icon-replacement grammar is based on a generalization of conventional string language concepts with attribute evaluation functions attached to its production rules. The picture processing grammar is a formal model for two-dimensional picture languages generalizing context-free grammars by allowing each object to have an arbitrary number of attributes. Relation grammars are a syntactic formalism for describing visual languages, where each sentence in a language is represented as a set of visual objects and a set of relationships among them. Finally, the unification-based grammar – a formalism in the field of linguistic information – is extended with functionally specified constraints in order to encompass visual languages whose objects are combined by operations rather than simple concatenation.

unified framework; visual languages; grammar based specification

nije evidentirano