Combining Case-Based and Model-Based Reasoning for Predicting the Outcome of Legal Cases

This paper presents an algorithm called IBP that combines case-based and model-based reasoning for an interpretive CBR application, predicting the outcome of legal cases. IBP uses a weak model of the domain to identify the issues raised in a case, and to combine the analyses for these issues; it reasons with cases to resolve conflicting evidence related to each issue. IBP reasons symbolically about the relevance of cases and uses evidential inferences. Experiments with a collection of historic cases show that IBP’s predictions are better than those made with its weak model or with cases alone. IBP also has higher accuracy compared to standard inductive and instance-based learning algorithms. 1 Reasoning with Cases for Prediction

• Type: Conference
• Year: 2003
• Where: ICCBR
• Authors: Stefanie Brüninghaus, Kevin D. Ashley
• Keywords: Algorithm Called Ibp, IBP’s Predictions, ICCBR 2003, Weak Model