ZoningLLM – A Novel Multimodal Application for Zoning Analysis

Abstract

Large municipalities in Canada have recently faced an unprecedented housing crisis. This has been driven by an increase in the demand for housing and a lack of housing supply. Stringent zoning requirements have contributed to reducing the construction of new housing. Each municipality typically has its own separate zoning code consisting of lengthy documents written in technical jargon. It is difficult for the public, researchers, and home builders alike to extract relevant information from these documents. This opacity restricts the discussion of zoning policy and aggravates the housing crisis. This project aimed to use generative and geomatic AI methods to analyze zoning and construction documents for the Waterloo Region to gain insights about zoning restrictions. This can be used to quantify and monitor the effects of zoning on housing supply. A webbased application with the ability to process, export and query ad-hoc zoning queries has been developed. Discussions with regional planners have underscored the importance of this work. Keywords: Large-Language Models, Zoning, Housing