Arrows are major components of diagrams, where they are typically used to facilitate the communication of spatial and temporal knowledge. An automated interpretation of arrow diagrams would be highly desirable in pen-based interfaces. This paper develops a method for deducing possible interpretations of arrow diagrams, which is composed of a uni-directional arrow symbol and one or more components. Based on a study of the use of arrow diagrams, we classify their semantics into properties, annotations, actions, and conjunctions. Then, we discuss the structural requirements of arrow diagrams for illustrating each class of semantics, as well as the structural rules for adding optional components. Finally, we investigate all possible structures of simple arrow diagrams for each class of semantics and demonstrate that knowledge about the structure of an arrow diagram reduces the ambiguity of its interpretation.
Yohei Kurata, Max J. Egenhofer