A condition for the solvability of the nonlinear model matching problem

Author(s):  
M. D. Di Benedetto
Author(s):  
R. Castro ◽  
M.D. Di Benedetto

2020 ◽  
Vol 12 (1) ◽  
pp. 1-19
Author(s):  
Mostefai Abdelkader ◽  
Ignacio García Rodríguez de Guzmán

This paper formulates the process model matching problem as an optimization problem and presents a heuristic approach based on genetic algorithms for computing a good enough alignment. An alignment is a set of not overlapping correspondences (i.e., pairs) between two process models(i.e., BP) and each correspondence is a pair of two sets of activities that represent the same behavior. The first set belongs to a source BP and the second set to a target BP. The proposed approach computes the solution by searching, over all possible alignments, the one that maximizes the intra-pairs cohesion while minimizing inter-pairs coupling. Cohesion of pairs and coupling between them is assessed using a proposed heuristic that combines syntactic and semantic similarity metrics. The proposed approach was evaluated on three well-known datasets. The results of the experiment showed that the approach has the potential to match business process models effectively.


Sign in / Sign up

Export Citation Format

Share Document