PhD thesis, pages 1-157. IFI, University of Zürich, October 2000. Available at http://www.ifi.unizh.ch/Forschung/dissertationen.html.
Abstract: The work elaborates the dichotomy of clustering techniques - which are a key feature for software engineering - to folding techniques - which preserve behaviour and enjoy strong theoretical properties - for both, Petri nets and Reverse Engineering. In a new framework of Petri Net categories this dichotomy shows up as adjunctions as does the interplay of structure and behaviour. Reachability and process semantics are attached categorically and a new variant of occurrence nets is proposed as a purer image of causality and branching. Different methods to translate Reverse Engineering problems to Petri Nets are proposed. A folding-based algorithm which transforms an unstructured flat net into a coloured net is introduced. In reverse engineering terms, it recovers a high-level design, a structured specification and a data model from an existing system. The algorithm allows adaptation to different tasks and its cost is almost linear, thus ensuring scalability.
Keywords: Petri nets; reverse software engineering; clustering; folding; structure; behaviour; semantics; occurrence systems; category theory; design metaphor.