Causal Prototype Libraries: Building and Retrieving Cases Based on Causal, Not Just Correlational, Structure

Most CBDS systems (e.g., Koornneef et al., 2021; Shved et al., 2024) retrieve cases based on feature similarity, but as Mannava (2024) argues, correlation is not causation, and decisions based on spurious similarities can mislead. The novel idea here is to construct case libraries where each case includes explicit causal graphs (e.g., using structural causal models) relating actions, contexts, and outcomes. Prototype retrieval then prioritizes not only surface similarity but also causal similarity—retrieving cases whose interventions led to outcomes through similar mechanisms. This approach could dramatically improve the reliability and explainability of recommendations, especially in safety-critical domains like healthcare or aviation. It also provides a new angle for research: how best to extract, represent, and query causal relationships in large, real-world case libraries. This causal turn in CBDS retrieval remains largely unexplored and could be transformative.

References:

1. A Web-Based Decision Support System for Aircraft Dispatch and Maintenance. H. Koornneef, W. Verhagen, R. Curran (2021). Aerospace.
2. INTELLECTUAL SUPPORT OF THE PROCESSES OF SEARCHING AND EXTRACTION OF PRECEDENTS IN CASE-BASED REASONING APPROACH. A. Shved, Ye. O. Davydenko, H. V. Horban (2024). Radio Electronics, Computer Science, Control.
3. Causal Inference in AI Based Decision Support: Beyond Correlation to Causation. Mohan K Mannava (2024). 2024 4th International Conference on Ubiquitous Computing and Intelligent Information Systems (ICUIS).