Communicating with Pictorial Databases

Abstract

This chapter presents a model and a language for specifying spatial queries: predicates involving the position on objects in a database picture. For example, “hospitals located within ten miles of a given school,” “cities adjacent to Los Angeles,” or “streets that intersect Highway 10 in downtown LA.” Most existing pictorial query languages are text oriented, e.g., GRAIN,⁽²⁾ IDMS, PSQL,⁽¹¹⁾ MIM,⁽¹³⁾ and PROBE.⁽¹⁰⁾ Others such as qpe⁽³⁾ and PICQUERY,⁽⁹⁾ follow the query-by-example approach (insertion of example values in tables), but do not make use of the visual properties of pictorial data. The language presented here differs from these methods by using pictures for specifying queries. These pictures, termed query pictures, visually depict conditions to be satisfied by the results of a query. For instance, Fig. la sketches a street intersection in which the condition “automobiles and people simultaneously entering a crosswalk” is satisfied. We enable new interaction mechanisms built upon the use of visual structures to manipulate spatially related objects. The work is related to visual programming languages as defined in Ref. 4. The following discussion addresses the problem of translating query pictures into first-order predicate calculus expressions describing queries. These expressions may be used by a query-processing mechanism to identify database pictures satisfying the desired conditions.