Incremental Interpretation of On-Line Hand-Drawn Structured Documents

This paper deals with the incremental interpretation of on-line hand-drawn structured documents. We recall the bases of the CD-CMG visual grammar we have presented previously, but we mainly focus on its associated parser. The formalism context-sensitiveness allows to exploit a parser with predictive capacities which couples bottomup and top-down strategies. It also helps targeting consistent candidate symbols, which significantly reduces the combinatory. We use the fuzzy set framework to model the adequacy measure of each possible interpretation and demonstrate how this permits to detect both uncertainty and the need for segmentation. Finally, in order to detect the necessity to wait before making a decision, we exploit partial interpretation evaluation. We present experiments highlighting the quickness of the analysis process, which makes it usable in real-time applications. This technology has been used to design the “Script&Go Electrical Sketches” software that is already on the market.