Active workspaces: distributed collaborative systems based on guarded attribute grammars

This paper presents a purely declarative approach to artifact-centric collaborative systems, a model which we introduce in two stages. First, we assume that the workspace of a user is given by a mindmap, shortened to a map, which is a tree used to visualize and organize tasks in which he or she is involved, with the information used for the resolution of these tasks. We introduce a model of guarded attribute grammar, or GAG, to help the automation of updating such a map. A GAG consists of an underlying grammar, that specifies the logical structure of the map, with semantic rules which are used both to govern the evolution of the tree structure (how an open node may be refined to a subtree) and to compute the value of some of its attributes (which derives from contextual information). The map enriched with this extra information is termed an active workspace. Second, we define collaborative systems by making the various user's active workspaces communicate with each other. The communication uses message passing without shared memory thus enabling convenient distribution on an asynchronous architecture. We present some formal properties of the model of guarded attribute grammars, then a language for their specification and we illustrate the approach on a case study for a disease surveillance system.