Video-Recorded Collaborative Annotation: Capturing the Legal Expertise of Multiple Annotators for Similar Case Matching

This thesis investigates the union between legal expertise and artificial intelligence, aiming to develop an explainable and reliable system for augmented reading and similar legal case matching. The research introduces a novel methodology designed to capture and formalize the implicit knowledge employed by legal professionals during case analysis. A core component of this methodology involves video-recorded annotations of legal experts engaged in the intricate processes of reading and comparing legal judgments. This technique provides deeper insights into the often tacit heuristics and decision-making processes inherent in legal reasoning, which are typically difficult to articulate explicitly. Careful analysis of these recordings allows for the extraction of valuable insights into the cognitive strategies employed by experts. The extracted knowledge is subsequently formalized into a computational pipeline that enables the automation of tasks such as identifying semantically similar cases, offering significant support to legal practitioners. By providing a structured approach to case analysis, this system has the potential to reduce backlogs in judicial systems, enhancing efficiency and decision-making speed. This thesis argues that grounding the development of AI systems in the actual practices of legal experts leads to the creation of more transparent, trustworthy, and ultimately more effective tools for navigating the complexities of legal reasoning. By doing so, it opens pathways for the employment of Artificial Intelligence to support legal processes, ensuring that technological advancements align closely with professional practices and ethical standards. Through a careful fusion of legal expertise and AI, this work aims to contribute to the optimization of legal practices.