I-Corps: Artificial intelligence-driven process for computationally predicting the outcomes of civil legal matters

这个i-Corps项目的更广泛的影响/商业潜力是开发一种人工智能驱动的过程，用于通过计算预测民事法律事务的结果。目前，律师对一项新的民事法律事项的可行性和价值进行评估。拟议的技术利用人工智能来标准化、系统化和外部化这一耗时且次优的过程。该平台接收新的案件信息，并生成对案件的评估和对其可能的和解价值的预测。这代表着通过改进案件选择、提高效率和更好的决策来管理法律事务的进步。这个i-Corps项目基于人工智能系统的开发，该系统提供民事法律案件的价值和针对分析采取的适当行动。提出的技术使用了一个建模过程，其中特征和相关权重在多个计算模型中组合，包括数据驱动的人工智能(AI)模型、基于规则的模型和受元分析研究约束的模型。此外，拟议的技术将包括一个使用自然语言处理的中央数据库，以将收集到的案例呈现为计算有用的形式。一个挑战是，这些数据通常是保密的，存储在不同的位置。将部署联合学习，这是一种在保存本地数据样本的多个分散数据库中训练算法而不交换数据样本的技术。该方法是首次将联合学习应用于法律数据，使产品能够扩展到众多用户，同时确保隐私和机密性。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this I-Corps project is the development of an artificial intelligence-driven process for computationally predicting the outcomes of civil legal matters. At present, attorneys estimate the viability and value of a new civil legal matter. The proposed technology leverages artificial intelligence to standardize, systematize, and externalize this time-intensive and suboptimal process. The platform intakes new case information and generate both an assessment of the case and a prediction of its likely settlement value. This represents an advance in managing legal affairs by improved case selection, increased efficiency, and better decision-making.This I-Corps project is based on the development of an artificial intelligence system that provides civil legal case values and the appropriate action to take in response to the analysis. The proposed technology uses a modeling process where features and associated weights are combined across multiple computational models, including data-driven artificial intelligence (AI) models, rule-based models, and models bounded by meta-analytic research. In addition, the proposed technology will include a centralized database using natural language processing to render the garnered cases into a computationally useful form. One challenge is that the data are often confidential and stored in disparate locations. Federated learning, a technique that trains an algorithm across multiple decentralized databases holding local data samples without exchanging the data samples, will be deployed. The method, which is the first application of federated learning to legal data, enables expansion of the product to numerous users while ensuring privacy and confidentiality.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.