| Description | Legal services address matters that are highly relevant to all citizens, e.g., about legal rights, obligations, and penalties concerning debt, housing, employment, personal life, and so on. Citizens Advice provides long list of such services (https://www.citizensadvice.org.uk/). Information online tends to use websites with text, which is often overwhelming, complex, and not personalised. There are advisors, but there are not enough of them with enough time. Online, interactive legal services (aka Expert Systems) would provide a way to offer accessible legal information to citizens that is tailored to them and their circumstances. However, creating such services has been very resource heavy, involving expertise about the domain along with programming skills.
To facilitate the creation of such system, recent advances have been made to use Large Language Models/GPTs to partially automate creation of the required templates that are then served online.
The project involves: identification and analysis of a legal service; analysis of a corpus of legal documents, specification and design; application of a tool to the corpus of documents; implementation in DocAssemble (Python, YAML, and Markdown); evaluation of the tool with the relevant user community. The result would progress development of a legal service that could useful to end users about particular legal problems.
This is a project written and supervised by Adam Wyner. It should not be used by other supervisors without permission. |
|---|