KONTERM: Exploratory Data Analysis for Semi-automatic Indexation of Legal Documents.
Erich Schweighofer, Werner Winiwarter · 1995
Exploratory data analysis aims at detecting structures and regularities in text. Based on the hypothesis of exploratory data analysis, we design our prototype KONTERM by representing legal knowledge using statistical techniques combined with intellectual revision. Previous research explored the feasibility of various statistical approaches also incorporating the application of neural networks [1, 2]. The major obstacle to traditional legal information systems is the difficult and often impossible handling of the particularities of legal language. The best solution seems to be the representation of legal knowledge by simple linguistic tools. Such an approach could also solve the bottleneck of knowledge acquisition for the creation of conceptual information retrieval systems. We compare an automatically produced document description by our prototype KONTERM with the intellectual indexation of the Swiss Federal Court. The results of KONTERM are paramount because KONTERM finds much more ...