Automatic layout in the face of unattached comments

Visual languages based on node-link diagrams are widely used for systems modeling. As in textual languages, comments can make diagrams easier to understand. In the absence of an explicit attachment between comments and the diagram elements they relate to, that relationship is usually given implicitly by the manual placement of comments near the related elements. While algorithms for the automatic layout of diagrams can make working with diagrams more effective, they usually fail to preserve implicit attachments by placing comments at arbitrary positions. In this paper, we propose a comment attachment algorithm that extracts implicit attachments and makes them accessible to layout algorithms. We implemented the algorithm in an application for browsing Ptolemy diagrams and achieved success rates, i. e. attachments as intended by the user, of up to 90 %.