Abstract—This paper presents a potential solution to the problem of extracting relevant sentences from past court decisions, which is an important first step of our legal deep learning research project. Court decisions are typically written in natural language like English. Hence, our extraction solution first uses legal statutes to construct an ontology for the desired sentences, and then uses NLTK (Natural Language Toolkit), a Python Natural Language Processing Toolkit, to construct search patterns based on the ontology to extract relevant passages from hundreds or thousands of past court decisions. The extracted sentences will be further processed and the resulting information will then be fed into a deep learning system, whose purpose is to assist legal practitioners by selecting relevant documents and streamline litigation.
Index Terms—Legal deep learning research project, ontology, nltk (natural language processing toolkit), tokens, stemmer, semantic similarity, wordnet.
Wai Yin Mok is with Department of Information Systems, The University of Alabama in Huntsville, Huntsville, Alabama, 35899, USA (e-mail: mokw@uah.edu).
Jonathan R. Mok was with School of Law, The University of Alabama Houston, Tuscaloosa, AL 35487, USA (e-mail: Jon.mok@law.ua.edu).
[PDF]
Cite: Wai Yin Mok and Jonathan R. Mok, "Extracting Relevant Sentences from Past Court Decisions: An Important First Step of A Legal Deep Learning Research Project," International Journal of Knowledge Engineering vol. 4, no. 1, pp. 17-20, 2018.