Title: A semantic abstraction criterion to reduce complexity on automatically discovered declarative maps
Authors: Pedro H. Piccoli Richetti; Fernanda Araujo Baião; Flávia Maria Santoro
Addresses: Federal University of Rio de Janeiro, Rio de Janeiro, Brazil ' Pontifical Catholic University of Rio de Janeiro, Rio de Janeiro, Brazil ' University of the State of Rio de Janeiro, Rio de Janeiro, Brazil
Abstract: A declarative approach can be employed on the definition of constraints that limit process execution possibilities. This perspective is appropriate when dealing with unstructured or flexible processes, a.k.a. knowledge-intensive processes. Declarative process mining may result in complex models due the discovery of a high quantity of constraints, producing models with excessive complexity. As abstractions are seen as an effective approach to represent readable models, this work proposes to create language-independent hierarchical declarative maps using a linguistic hierarchy of activities. The proposed approach applies natural language processing to build more abstract declarative models produced by process mining. The presented method was evaluated in a case study with real life data and support from domain experts. The findings showed that it is possible to generate meaningful groups by looking for the semantics of activity labels in order to create abstract process views with reduced complexity, starting from a low-level declarative map.
Keywords: process mining; declarative modelling; semantic abstraction; business processes; natural language processing; NTL.
International Journal of Business Process Integration and Management, 2022 Vol.11 No.1, pp.1 - 18
Accepted: 01 Nov 2019
Published online: 01 Sep 2022 *