The Squash Behavior Graph Routing between the Business Process Event Capsules

• Published in: 2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT) pp. 1 - 9
• Main Authors: Chaima, Afifi, Ali, Khebizi, Khaled, Halimi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 06.07.2023
• Subjects: Behavior Graph; Business Process Graph; Computer architecture; Data models; Event-transition; Feature extraction; Graph-Capsule Neural nets; Neural networks; Predictive models; Routing; Semantics
• ISSN: 2473-7674
• DOI: 10.1109/ICCCNT56998.2023.10307176

In recent years, a set of Process Mining (PM) techniques and approaches has been developed to extract knowledge's from business processes (BP) execution data. Although, these conventional PM techniques are pretty efficient to extract handcrafted features from the event log data, they still lack of the automatic features extraction that Capsule-Neural-Nets (Capsnet) models processes from the graph. These caps-net models express the node features of the constructed graph in form of capsules, and, the relevant insights at the graph level are captured by deploying the routing mechanism.This paper, proposes a novel Squash Behavior Graph Routing (SGBR) mechanism that constitute the backbone foundation of an Hierarchical Caps-Graph Neural-Net (H-Caps-GNN) architecture. The routing based features propagation mechanism in H-Caps-GNN structure is able to learn unified events feature's of BP and their relationships. Those discovered properties represents the valuable insights for automatically learning the semantic of BP features which can be used in various BP applications' (Process discovery, conformance checking, event prediction,...etc)