Privé : Procezeus – AI powered chatbot for landlords and tenants disputes

The team

The team of Capstone is composed of students from Concordia University and Université de Montréal. Mihai Damaschin, Lance Lafontaine, Arek Manoukian, Bruce Zhipeng Cai, Samuel Campbell , Sylvain Czyzewski, Taimoor Rana and Nareg Froundjian worked together in collaboration with the Cyberjustice Laboratory.



The Process

The goal of Procezeus is to use a natural language and automated modeling to provide a system capable of dialoguing with users and respond to their legal enquiries, however formulated, by referring them to frequently asked questions, similar cases, defining legal concepts and presenting their available legal avenues. The system would draw information by reading and analyzing legal documents, such as statutes and case law.


For example, the prolonging and ineffectiveness of landlord-tenant dispute resolutions, both judicially and extra judicially, forces the victims to remain in precarious living situations often in violation of the law. Automated, high-speed reading of decisions and harvesting the rules that underlie them would allow the bot to directly search the information required by the user and to conduct its line of questioning accordingly. By basing its answers on case law, the bot does not need human intervention to keep up with jurisprudential developments. Users will have the ability to converse with the chatbot, describing in detail the situation which they wish to seek guidance for. With the help of RasaAI, the system is able to understand a user’s dilemma and ask a series of follow-up questions that allows for the creation of a case profile. The case profile is then processed by the machine learning service to obtain a series of possible outcomes for the case. A person experiencing such legal issues would be able to know the available options in their situation without having to consult legal professionals. This information would be provided in a simple and understandable way, contrarily to the legal documentation the user would have otherwise access to.



Ce contenu a été mis à jour le 26 juin 2018 à 13 h 53 min.