Conversational Agents in the Legal Domain: A Systematic Review of the Literature

Legal documents and norms can be truly complicated. This is not news, but it means that much helpful information and many resources are currently inaccessible to the vast majority of people without the aid of a qualified individual. Chatbots, and more generally conversational agents, can provide assistance and support to users for this task by enabling a dialogue in a natural and colloquial language and providing intelligent answers based on the domain-specific knowledge available. This study sought to explore the state of the art in implementing a conversational system in the legal domain focusing on the main technologies adopted to tackle issues that arise. The search was conducted on four different databases: ACM Digital Library, IEEE Xplore, Scopus, Web of Science. A total of 1.532 articles were extracted, of which 22 met both the inclusion and exclusion criteria during the screening phase. The results revealed many natural language processing techniques that can enhance conversational agents in the interactions with users’ questions concerning the legal domain. Our article pays particular attention to the underrepresented implementative aspects of such systems and individuates a literature gap in the majority of the solutions proposed that consisted only of a retrieval approach. While retrieval-based systems can efficiently locate and present relevant information from predefined datasets, they often lack the ability to generate the nuanced, context-sensitive responses that are crucial in legal settings. This limitation influences the effectiveness of these agents, as legal inquiries frequently require not just the retrieval of information, but also the ability to interpret and rephrase complex legal language into more understandable terms for the user. These findings pave the way for further research on the subject treated that embrace techniques such as retrieval-augmented generation and explore more suitable approaches for evaluating these systems.