Empowering Prior to Court Legal Analysis Dataset

本研究引入了一个名为“Empowering Prior to Court Legal Analysis Dataset”的新数据集，由金斯顿大学伦敦创建，包含687条警方审讯中的陈述记录。数据集通过精细筛选，去除了无关元素，如填充词和非言语行为指示，专注于言语真实性线索的分析。该数据集旨在支持自然语言处理和法律信息学领域中陈述分类的研究，特别关注区分真实与欺骗性陈述。通过此数据集，研究者开发了精细调整的DistilBERT模型，以提高分类准确性，并通过可解释人工智能技术增强模型的透明度和可信度。该数据集的应用领域主要集中在法律实践和研究中，旨在通过自动化工具提高法律决策的效率和公正性。

This study introduces a new dataset named "Empowering Prior to Court Legal Analysis Dataset", created by Kingston University London. The dataset contains 687 recorded statements from police interrogations. After rigorous filtering to remove irrelevant elements such as filler words and non-verbal behavior cues, it focuses on the analysis of verbal truthfulness cues. It aims to support research on statement classification in the fields of Natural Language Processing (NLP) and Legal Informatics, with a particular focus on distinguishing between truthful and deceptive statements. Leveraging this dataset, researchers developed fine-tuned DistilBERT models to improve classification accuracy, and enhanced the model's transparency and credibility through Explainable Artificial Intelligence (XAI) techniques. The dataset is primarily applied in legal practice and research, aiming to improve the efficiency and fairness of legal decision-making via automated tools.