Department of Software Technology
Vienna University of Technology


The Exploration of Legal Text Corpora with Hierarchical Neural Networks: A Guided Tour in Public International Law The classification of feature vectors representing the interpretation of legal documents improves the search for similar or related documents, the interpretation of these documents as well as the navigation within the text corpus. The need for effective approaches of classification is dramatically increased nowadays due to the advent of massive digital libraries containing free-form legal text documents. What we are looking for are powerful methods for the exploration of such libraries whereby the detection of similarities between groups of documents is the overall goal. In other words, methods that may be used to gain insight in the inherent structure of the various items contained in a text archive are needed. In this paper we present the results from a case study in legal document classification based on an experimental document archive comprising important treaties in public international law. The core task of classification is performed by a non-standard neural network model with a layered architecture consisting of mutually independent unsupervised neural networks. The distinguished features of this learning architecture is the remarkably fast training time combined with the benefit of explicit cluster representation. The access to legal text archives may be enhanced by guided tours providing the means for convenient voyage in an environment of dynamically classified legal documents.


Up

Comments: rauber@ifs.tuwien.ac.at