Testing Shape Descriptions by Automatically Translating them for Use in Sketch Recognition

What: We present a translator which takes symbolic descriptions of how shapes are drawn, displayed, and edited in a domain and automatically translates them into shape recognizers, editing recognizers, and shape exhibitors for use in our domain independent sketch recognition system. The implementation of this translator and domain independent sketch recognition system serves to show both that such a framework is feasible and that LADDER [1] is an acceptable language for describing domain information. The aim of our implementation is its simplicity in design. Why: To date, sketch recognition systems have been domain-specific, with the recognition details of the domain hard-coded into the system. Developing such a sketch interface is a substantial effort. We propose that rather than build a separate recognition system for each domain, we instead build a single domain independent recognition system that can be customized for multiple domains. To build a sketch recognition system for a new domain, the developer would need only to write a domain description, describing how domain shapes are drawn, displayed and edited, for use with our domain independent sketch recognition system. This domain description would then be translated into shape recognizers, editing recognizers, and shape exhibitors for use in a domain independent sketch recognition system. The inspiration from such a framework stems from the speech recognition domain which has been using this approach with some success [2].