Knowledge-Based Cause-Effect Analysis Enriched by Generating Multi-Layered DSS Models

摘要

Computer-based analysis to support decision-making in organizations is a crucial competitive factor. Cause-effect analysis is an important component of these analyses, as it identifies cause-effect relationships amongst data, which can be applied in decision-making situations to improve the decision-making quality. This paper envisions a concept for the support of cause-effect analyses, which is based on an integrated knowledge base with cause-effect relationships and a knowledge reasoning process according to the human approach to solving problems. The knowledge base integrates both structured and unstructured knowledge from a variety of organizational sources. The knowledge reasoning is divided into three phases during which the decision situation is (1) isolated and matched into the knowledge base, (2) explored for potential causes (including their validation) and finally (3) verified, and, if necessary, adjusted by the user. As a proof of concept, this concept is applied manually to the slightly extended example data set from Microsoft for the SQL Server 2012. For the creation of the knowledge base, knowledge about the cause-effect relationships is extracted manually from the database schemas and integrated with additional expert knowledge about further cause-effect relationships. The result is an ontology with cause-effect relationships for this specific data set. Based on a fictitious decision scenario the phases of the knowledge reasoning are played through. The exploration of the ontology will typically identify cause-effect chains with various potential explanations alongside the levels of the chain. These potential cause-effect chains are implemented in a DSS model with multiple layers. The resulting DSS model enables the evaluation of the impact of the identified cause-effect chains for the specific decision scenario.