On Comments in Visual Languages
Visual languages based on node-link diagrams can be used to develop software and, like textual languages, offer the possibility to write explanatory comments.
Which node a comment refers to is usually not made explicit, but is implicitly clear to readers through placement and content. While automatic layout algorithms can make working with diagrams more productive, they tend to destroy such implicit clues because they are not aware of them and thus do not preserve the relative placement of comments and the nodes they refer to. Implicit clues thus need to be inferred and made explicit to be taken into account by layout algorithms. This is what we call the comment attachment problem.
In this paper, we improve upon a previous paper on the subject , introducing further heuristics that aim to describe relations between comments and nodes.
Based on an analysis of comment placement in a set of example diagrams, we develop a general comment attachment framework and evaluate the quality of its inferred attachments.