%0 Conference Proceedings
%T Mind the Gap!: Learning Missing Constraints from Annotated Conceptual Model Simulations
%+ Free University of Bozen-Bolzano
%+ Services, Cybersecurity and Safety (SCS)
%A Fumagalli, Mattia
%A Sales, Tiago, Prince
%A Guizzardi, Giancarlo
%Z Part 1: Enterprise Modeling and Enterprise Architecture
%< avec comité de lecture
%@ 978-3-030-91278-9
%( Lecture Notes in Business Information Processing
%B 14th IFIP Working Conference on The Practice of Enterprise Modeling (PoEM)
%C Riga, Latvia
%Y Estefanía Serral
%Y Janis Stirna
%Y Jolita Ralyté
%Y Jānis Grabis
%I Springer International Publishing
%3 The Practice of Enterprise Modeling
%V LNBIP-432
%P 64-79
%8 2021-11-24
%D 2021
%R 10.1007/978-3-030-91279-6_5
%K Conceptual modeling
%K Model validation
%K Inductive learning
%K Model simulation
%Z Computer Science [cs]
%Z Humanities and Social Sciences/Library and information sciencesConference papers
%X Conceptual modeling plays a fundamental role to capture information about complex business domains (e.g., finance, healthcare) and enables semantic interoperability. To fulfill their role, conceptual models must contain the exact set of constraints that represent the worldview of the relevant domain stakeholders. However, as empirical results show, modelers are subject to cognitive limitations and biases and, hence, in practice, they produce models that fall short in that respect. Moreover, the process of formally designing conceptual models is notoriously hard and requires expertise that modelers do not always have. This paper falls in the general area concerned with the development of artificial intelligence techniques for the enterprise. In particular, we propose an approach that leverages model finding and inductive logic programming (ILP) techniques. We aim to move towards supporting modelers in identifying domain constraints that are missing from their models, and thus improving their precision w.r.t. their intended worldviews. Firstly, we describe how to use the results produced by the application of model finding as input to an inductive learning process. Secondly, we test the approach with the goal of demonstrating its feasibility and illustrating some key design issues to be considered while using these techniques.
%G English
%Z TC 8
%Z WG 8.1
%2 https://inria.hal.science/hal-04323862/document
%2 https://inria.hal.science/hal-04323862/file/514409_1_En_5_Chapter.pdf
%L hal-04323862
%U https://inria.hal.science/hal-04323862
%~ SHS
%~ IFIP
%~ IFIP-TC
%~ IFIP-LNBIP
%~ IFIP-WG
%~ IFIP-TC8
%~ IFIP-WG8-1
%~ IFIP-POEM
%~ IFIP-LNBIP-432