A novel decision graph model to visualise dominance based if-then decision rules

When decision makers rely on large sets of if-then decision rules to guide complex choices, information overload quickly becomes a barrier to effective use.The Dominance-based Rough Set Approach (DRSA) is a widely used preference learning method whose primary output is a set of if-then decision rules incorporating preference orders over attributes and decision classes.Although effective for knowledge representation and classification, large rule sets can overwhelm decision makers and hinder interpretation.To address this challenge, this paper proposes a new decision graph model for visualising dominance based if-then decision rules.The construction process begins with a preprocessing phase that extracts a binary decision matrix from rule descriptions.To avoid generating nonexistent rules, this matrix may require transformation through column splitting or a combination of column splitting and ordering-based strategies.The transformed matrix is then used to build a directed acyclic decision graph in which nodes represent elementary conditions and decisions, and edges denote their logical relationships.Each path from the root to a decision node corresponds to a complete decision rule.The paper introduces the core concepts of decision graphs and presents algorithms for their construction, linearisation, and exploitation.Both linearisation and exploitation algorithms are based on the Breadth-First Search graph traversal algorithm.The approach is demonstrated using real-world cultural heritage data from Europe, and expert feedback indicates that decision graphs reduce cognitive effort, improve communication, and provide a clear and accessible complement to traditional textual and tabular representations of decision rules.