Abstract:In common law jurisdictions, the notion of proof beyond a reasonable doubt is frequently related to notions such as the belief or certainty of a judge or a juror about reality. The notion of balance of probabilities is however related to likelihood or probability. In the present study, we link belief and proof by introducing the notion of epistemic modality, which is concerned with the speaker's belief in propositional probability. The variation in the orientation of epistemic modality helps to integrate the two levels of proof and bridge the apparent test gap between them. The notion of relevance is further introduced in order to clarify the nature of legal proof by taking rape cases as example. This study also provides an integrated model to improve but diversify the expressions in terms of the burden of proof. For most courts, court judgments are processed only according to the general case data, procedural context; such kinds of fact-based information processing and information retrieval seldom help the court to make its decision unless with tremendous and repetitious work. For the consistency and efficiency of court adjudication, it is suggested in the present study that a Knowledge Management (KM) model mainly based on elements and factors which decide or affect the criminal liability. Such a KM model provides an overall framework, though non-exhaustive, and therefore makes court adjudication within narrow discretion and achieves the maximum justice.

Language Modeling for Legal Proof

Gauging Court Adjudication: Qualification And Quantification

Epistemic modality in court judgments: A corpus-driven comparison of civil cases in Hong Kong and Scotland

Are LLMs Rigorous Logical Reasoner? Empowering Natural Language Proof Generation with Contrastive Stepwise Decoding

Can Large Language Models Grasp Legal Theories? Enhance Legal Reasoning with Insights from Multi-Agent Collaboration

Evaluating statistical language models as pragmatic reasoners

Justifiable Artificial Intelligence: Engineering Large Language Models for Legal Applications

Language Models with Conformal Factuality Guarantees

The Factuality of Large Language Models in the Legal Domain

What Are the Odds? Language Models Are Capable of Probabilistic Reasoning

Does Reasoning Emerge? Examining the Probabilities of Causation in Large Language Models

Probing neural language models for understanding of words of estimative probability

Large Language Models are legal but they are not: Making the case for a powerful LegalLLM

Enhancing Logical Reasoning in Large Language Models to Facilitate Legal Applications

Large Language Models and Explainable Law: a Hybrid Methodology

Language Models Are Greedy Reasoners: A Systematic Formal Analysis of Chain-of-Thought

Towards Large Language Models as Copilots for Theorem Proving in Lean

Understanding Epistemic Language with a Bayesian Theory of Mind

Alignment Between the Decision-Making Logic of LLMs and Human Cognition: A Case Study on Legal LLMs

CDMBE: A Case Description Model Based on Evidence

Large language models as tax attorneys: a case study in legal capabilities emergence