Abstract
Problems. An applied problem of automating the search for relevant legal positions on a set of court decisions in the judiciary of Ukraine. The possibility of developing a high-quality NLP system for searching for a relevant, contextually justified legal position on a set of court decisions in the judiciary of Ukraine, which, based on the user's request, produces expert answers to the user's request by applying a reasoning-search algorithm based on large language models. Objective. To develop and formalize the concept and conceptual model of the NLP system for finding legal positions on the basis of interdisciplinary knowledge of a technical and philosophical direction, which covers the existence of a person, community, enterprises, institutions and organizations, state authorities, and also substantiates the formation of a legal position on the basis of a plurality . legislative acts and court decisions. Methodology of implementation. The method of implementation of the NLP system for searching for a context-based legal position lies in the interaction of the essences of tasks related to the judiciary of Ukraine, tools for analysis, processing, text generation and available available court decisions of the judiciary of Ukraine based on the Erikson-Penker business profile and large language models . Results. The proposed interdisciplinary model of the concept of the NLP system of finding a solution to the problem, tasks and tasks of searching court documents for more rendering of court decisions and/or formation of a legal position in the form of a Venn diagram as a set of entities and their cooperation, which integrates philosophical, legal and technical knowledge. The Erikson-Penker business profile of the conceptual model of the NLP system for searching for a context-based legal position is formalized. A formalized model of the work process of the NLP system of searching for a context-based legal position based on the RAG method. A conceptual model of the NLP system for searching for a context-based legal position has been developed in the form of a structural mapping based on a component diagram and a dynamic mapping based on an activity diagram with a description of the internal structure of the components. Conclusions. A number of important philosophical, legal and technical problems have been identified in the process of searching for legal positions on many court decisions. The interdisciplinary nature of the problem of developing the concept and conceptual model of the NLP system for finding a context-based legal position on a set of court decisions and the need to automate the process of finding a legal position for judicial practice has been established. At the meta levels, the concept of the NLP system for finding context-based legal positions on Venn diagrams and the conceptual model of the NLP system based on the modified Erikson-Penker business profile as a system of entities and cooperation are formalized. A conceptual model of the NLP system for finding a context-based legal position has been developed in the form of a structural mapping based on a diagram of components and a dynamic mapping based on an activity diagram with a description of the internal structure of the components. The functionality of the NLP system was compared with other main existing systems for searching court decisions in the judiciary and it was established that the developed system, at least, is not implemented in terms of functionality of existing systems and has additional opportunities for users.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call