2020 | OriginalPaper | Buchkapitel
Natural Language Processing of Russian Court Decisions for Digital Indicators Mapping for Oversight Process Control Efficiency: Disobeying a Police Officer Case
Autoren: Oleg Metsker, Egor Trofimov, Sofia Grechishcheva
Abstract
This article describes the study results in the development of the method of natural language processing (NLP) of semi-structured Russian court decisions to improve the quality of knowledge extraction describing legal process. Improving the accuracy of information retrieval from electronic records of court decisions was achieved with using combination of TF-IDF and latent semantic analysis. As a result, the word combinations of facts of offenses and procedural facts that may affect the decision-making of the court are identified. The applicability of the results is shown on the example of development a decision tree ML model of the appointment of arrest or fine punishment if disobeying a police officer. Automated mapping of court decisions texts on Russian language is also possible use for the development of artificial intelligence systems and new generation decision support systems in law domain.
