Predicting Exceptions in Agent-Based Supply-Chains

Abstract. Exceptions occur frequently in supply-chains. Consequently, detecting exceptions timely is of great practical value. To enable a timely detection of exceptions, this paper develops an agent-based model for supply-chains. The participants are modeled as autonomous agents that can reason about their interactions and communicate with other agents. The communications of the agents create and manipulate commitments. Violation of commitments leads to exceptions. We provide two methods for detecting such violations. First one uses an AND/OR tree to break down a commitment into smaller commitments and to decide on violation when sub parts of a commitment are violated. The second one uses an ontology to predict if a commitment is going to be violated. Using both approaches jointly, we can detect exceptions in a timely manner.

• Type: Conference
• Year: 2005
• Where: ESAW
• Authors: Albert Özkohen, Pinar Yolum
• Keywords: ESAW 2005, Great Practical Value, Timely, Timely Detection