Deep Graph Representation Learning for Business Process Modeling

EasyChair Preprint 9340

Abstract

Business process (BP) models can quickly become complex and expensive. In turn, the abstraction

has proved to be a challenging key for establishing a comprehensible and high-level view of the BP model.

Where the aggregated processes are preserved and irrelevant details are omitted. The promising research

question explores the reasonable stones on merging and validating the produced high-level model. The

semantic BP logic in its turn, is a cornerstone of extra-knowledge that contributes in the development of

the ideal BP high abstraction model.

This study focuses on the BP abstraction problem. Furthermore, with the remarkable development in ar

tificial intelligence (AI) techniques in the context of business process mining, BP models can be retrieved

from execution data utilizing deep learning (DL) approaches in general, and Deep Graph Representa

tion Learning (DGRL) in particular. This study emphasizes the unavailability of a DGRL model that

generates a BP model from execution traces. Finally, a roadmap for future research directions is proposed.

Keyphrases: Deep Graph Representation Learning, abstraction level, business process, graph theory, graph-based modeling

@booklet{EasyChair:9340,
  author    = {Chaima Afifi and Ali Khebizi},
  title     = {Deep Graph Representation Learning for Business Process Modeling},
  howpublished = {EasyChair Preprint 9340},
  year      = {EasyChair, 2022}}