IEEE Transactions on Pattern Analysis and Machine Intelligence
The complexity of word problems—this time with interleaving
Information and Computation
Inference of Reversible Languages
Journal of the ACM (JACM)
Shuffle languages, Petri nets, and context-sensitive grammars
Communications of the ACM
A context driven approach for workflow mining
IJCAI'09 Proceedings of the 21st international jont conference on Artifical intelligence
Design and control of workflow processes: business process management for the service industry
Design and control of workflow processes: business process management for the service industry
Hi-index | 0.00 |
Workflows are an important knowledge representation used to understand and automate processes in diverse task domains. Past work has explored the problem of learning workflows from traces of processing. In this paper, we are concerned with learning workflows from interleaved traces captured during the concurrent processing of multiple task instances. We first present an abstraction of the problem of recovering workflows from interleaved example traces in terms of grammar induction.We then describe a two-stage approach to reasoning about the problem, highlighting some negative results that demonstrate the need to work with a restricted class of languages. Finally, we give an example of a restricted language class called terminated languages for which an accepting deterministic finite automaton (DFA) can be recovered in the limit from interleaved strings, and make remarks about the applicability of the two-stage approach to terminated languages.