Neuro-fuzzy in legal reasoning

Trains a neuro-fuzzy system using both rule-based knowledge and inductive learning to find structure in legal precedent decisions as well as to identify legal precedents. Similar to humans, an information processing system should be able to exploit knowledge that is presented in form of rules as well as information that is acquired through experience. The author demonstrates how fuzzy rule-based knowledge can be used to pre-structure a neural network. In this way, the network has problem specific knowledge prior to training. After training, the altered fuzzy rules can be extracted and interpreted by an expert. The viability of the approach is demonstrated in a legal application, where fuzzy rules defined by a legal expert as well as previous court decisions are used for network structuring and training