Large language models (LLMs) and retrieval-augmented generation (RAG) are increasingly used in legal decision support, but retrieved evidence and fluent explanations do not guarantee valid normative inference. This paper proposes a proof-carrying neuro-symbolic method for non-monotonic legal reasoning. The LLM component is restricted to source-linked extraction of facts, defeasible rules, defeaters, priorities, citations, and operational confidence scores, while a deterministic symbolic engine computes the conclusion. Evidence is represented as a finite defeasible normative theory and compiled into a Dung-style argumentation framework; accepted conclusions are obtained from the grounded extension and returned with proof graphs showing support, attacks, and priority-based defeats. Under gold formalization, the symbolic engine achieved 99.3% accuracy on a 600-case controlled benchmark. In a 240-scenario LLM-to-logic experiment, the GPT-4o extractor followed by symbolic reasoning achieved 86.7% downstream accuracy versus 75.8% for a direct LLM over the same retrieved evidence; the paired difference was supported by an exact McNemar test after Holm correction (adjusted p = 0.016). Differences from the PDL and simpler symbolic baselines were not statistically established. Validation-triggered repair yielded 90.4% observed accuracy. Public-contract, Russian-law, stress-test, scalability, and lawyer-verification experiments further delimit the feasibility and current limitations of proof-carrying legal decision support.
Maxim Ulizko, Tatiana Polevaya, I. Tomilov et al.· Big Data and Cognitive Compu...· 0 citations
Analysis of document corpora with complex internal and inter-document links requires not only retrieval of relevant texts but also construction of a compact, verifiable and traceable set of fragments sufficient for downstream reasoning and citation. The purpose of this work is to propose a fragment retrieval method for document corpora in which the meaning of a fragment is determined by its local content, position in the document hierarchy and links to other fragments. The method is based on a joint representation of the corpus as a tree of structural units and a directed link graph as well as on hybrid ranking that combines lexical search, vector similarity, and a link signal. The monotonicity and submodularity of the objective function are shown, which makes it possible to use greedy algorithms with a known approximation guarantee and to perform budgeted context selection for a Retrieval-Augmented Generation (RAG) system. In addition, an evaluation protocol is introduced that separates retrieval quality at the document, fragment, and citation levels. The method is formally validated on tests of lexical mismatch robustness and budgeted selection efficiency compared with simple strategies. Examples from the legal domain are used to illustrate the method. The method can be used as a retrieval layer for RAG systems in question answering, evidence retrieval, regulatory compliance, and analysis of large structurally connected corpora.
M. Ulizko, A. Beresnev, V. V. Zhukov et al.· Scientific and Technical Jou...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
