Towards a Methodology for Comparing Legal Texts Based on Semantic, Storytelling and Natural Language Processing

Applications of artificial intelligence to text processing and manipulation have recently increased, and promising results are being achieved. Because of this confidence in these results, it would be an attractive challenge to harness the potential of natural language processing and semantic for text comparison tasks. This article operates in this context, taking the field of law as a scope and outlining the preliminary stages of research activity, which examines the possibility of finding a method for comparing texts from legal sources in order to reveal possible contradictions. The study that this paper discusses focuses on the following activities: uniform semantic representation of information of concern extracted from a text using an ontological storytelling model; evaluation of natural language processing techniques to retrieve such information within written documents; and finally, a system of logical inference rules to carry out fact-based comparisons to highlight discrepancies or alignments. The application case examined is in the legal domain, more precisely, it aims at the comparison of testimonies that tell different versions of the same fact.