Founded in 2009 by Professor Vern R. Walker, the LLT Lab conducts empirical research on argument and reasoning patterns in diverse samples of legal documents, creating the semantic data, training protocols and software analytics needed to develop technology to assist legal reasoning.
For the HOFSTRA LAW REPORT, Walker, who retired in 2017 but continues to direct the LLT Lab as professor emeritus, highlights the LLT Lab’s current and upcoming projects and collaborations.
LLT Lab Projects
The LLT Lab has three main projects: Vaccine/Injury Compensation, Medical Malpractice Compensation, and Veterans’ Claims. Each project involves:
• creating “semantic data” in the form of annotated adjudicatory decisions, which the LLT Lab and other collaborators can use for developing software to assist in identifying and extracting information from legal documents;
• creating the annotation protocols that we use to train new annotators and to check the accuracy of our work, as well as to obtain insights into developing rule-based software programs and features for machine learning; and
• identifying argument structures that are either generic to legal reasoning or specific to the reasoning found within the particular area of adjudicatory decision-making, and which can be used to guide argument generation in new cases.
University of Pittsburgh and Carnegie Mellon University
We continue to collaborate with the University of Pittsburgh’s Intelligent Systems Program, as well as Pitt’s Learning Research and Development Center, and Carnegie Mellon University’s Language Technologies Institute in the School of Computer Science on a project for automating argument mining using the LLT Lab’s veterans’ claims data and annotation protocols.
In the process, we help support Carnegie Mellon students in a capstone master’s course, where experimental computer programs are written and tested.
Hofstra Law’s Gitenstein Institute for Health Law and Policy
We continue to collaborate with the Gitenstein Institute to apply our developed tools to administrative and judicial decisions on veterans’ claims. We hope that the resulting ability to conduct “smart searches” for arguments and reasoning patterns over a data set of decided veterans’ cases will provide a resource for veterans and their representatives, as well as a demonstration of how to increase access to justice through a publicly available website and service.
We are collaborating not only on useful tools for students in Hofstra Law’s Robert W. Entenmann Veterans Law Clinic but also on tools for a planned “virtual” veterans’ clinic, which would allow veterans access online from anywhere.
Hofstra University Digital Research Center
In collaboration with the Hofstra University Digital Research Center, the LLT Lab is developing an online library of different reasoning structures that we find in legal texts. We are using the LLT Lab’s data sets to develop a library of legal texts that are annotated as examples of legal reasoning for use in undergraduate and graduate logic courses. We use Hofstra University’s instance of Annotation Studio, a software-annotation platform developed by the Massachusetts Institute of Technology. The next step is to create more content and a user-friendly website for the public to access the library.
University of Montreal’s Cyberjustice Laboratory
We have entered into a new major collaboration with the Cyberjustice Laboratory at the University of Montreal. The goal of the joint pilot project is to produce, within about one year, a fully functional prototype of a system that includes:
• at least two web applications to apply past decisions to new cases (one for our Veterans’ Claims project and one in French for landlord-tenant decisions in the Province of Quebec);
• a database format designed specifically to provide content for these web applications; and
• database content that will be created initially by manually extracting information on legal reasoning from two samples of the relevant legal decisions. We are using our decisions by the Board of Veterans’ Appeals (BVA); the Cyberjustice Laboratory is using decisions of the Régie du logement du Québec, the provincial government agency that oversees the residential rental market.
This joint project represents the first time that our methodology involving rule trees and semantic data will be applied in a language other than English and in a legal system based on the Napoleonic Code.
Particularly in connection with the project with the Cyberjustice Laboratory, but also more generally, we have begun active collaboration with the Department of Computer Science at Hofstra’s Fred DeMatteis School of Engineering and Applied Science (SEAS). It is excellent having another Hofstra team engaged with our Veterans’ Claims project.
During the next year, we will continue to focus our efforts on developing the Veterans’ Claims project. We will do this by working closely with our collaborators and developing data and tools that they find useful.
View more information about the LLT Lab on its website.