Universal Legal Article Prediction via Tight Collaboration between Supervised Classification Model and LLM

Legal Article Prediction (LAP) is a critical task in legal text classification, leveraging natural language processing (NLP) techniques to automatically predict relevant legal articles based on the fact descriptions of cases. As a foundational step in legal decision-making, LAP plays a pivotal role in determining subsequent judgments, such as charges and penalties. Despite its importance, existing methods face significant challenges in addressing the complexities of LAP. Supervised classification models (SCMs), such as CNN and BERT, struggle to fully capture intricate fact patterns due to their inherent limitations. Conversely, large language models (LLMs), while excelling in generative tasks, perform suboptimally in predictive scenarios due to the abstract and ID-based nature of legal articles. Furthermore, the diversity of legal systems across jurisdictions exacerbates the issue, as most approaches are tailored to specific countries and lack broader applicability. To address these limitations, we propose Uni-LAP, a universal framework for legal article prediction that integrates the strengths of SCMs and LLMs through tight collaboration. Specifically, in Uni-LAP, the SCM is enhanced with a novel Top-K loss function to generate accurate candidate articles, while the LLM employs syllogism-inspired reasoning to refine the final predictions. We evaluated Uni-LAP on datasets from multiple jurisdictions, and empirical results demonstrate that our approach consistently outperforms existing baselines, showcasing its effectiveness and generalizability.

Dispute resolution in legal mediation with quantitative argumentation

Mediation is often treated as an extension of negotiation, without taking into account the unique role that norms and facts play in legal mediation. Additionally, current approaches for updating argument acceptability in response to changing variables frequently require the introduction of new arguments or the removal of existing ones, which can be inefficient and cumbersome in decision-making processes within legal disputes. In this paper, our contribution is two-fold. First, we introduce a QuAM (Quantitative Argumentation Mediate) framework, which integrates the parties' knowledge and the mediator's knowledge, including facts and legal norms, when determining the acceptability of a mediation goal. Second, we develop a new formalism to model the relationship between the acceptability of a goal argument and the values assigned to a variable associated with the argument. We use a real-world legal mediation as a running example to illustrate our approach.

A Filtering-based General Approach to Learning Rational Constraints of Epistemic Graphs

Epistemic graphs generalize the epistemic approach to probabilistic argumentation and tackle the uncertainties in and between arguments. A framework was proposed to generate epistemic constraints from data using a two-way generalization method in the perspective of only considering the beliefs of participants without considering the nature of relations represented in an epistemic graph. The deficiency of original framework is that it is unable to learn rules using tighter constraints, and the learnt rules might be counterintuitive. Meanwhile, when dealing with more restricted values, the filtering computational complexity will increase sharply, and the time performance would become unreasonable. This paper introduces a filtering-based approach using a multiple-way generalization step to generate a set of rational rules based on both the beliefs of each agent on different arguments and the epistemic graph corresponding to the epistemic constraints. This approach is able to generated rational rules with multiple restricted values in higher efficiency. Meanwhile, we have proposed a standard to analyze the rationality of a dataset based on the postulates of deciding rational rules. We evaluate the filtering-based approach on two suitable data bases. The empirical results show that the filtering-based approach performs well with a better efficiency comparing to the original framework, and rules generated from the improved approach are ensured to be rational.