Evaluating Public Consultation in Urban Planning via Neural Language Models and Topic Modelling

Urban planning has the fundamental role of managing the cooperative development of ever bigger cities and the community's cultures inhabiting those places. To make the best decisions, urban planners need to analyse relevant data and community responses, spread through abundant reports. This process can be labour intensive and might lead to overlooking less prominent but still relevant concerns of minorities. In this study, we present a Natural Language Processing framework to assist the analysis of consultation reports. The framework leverages state-of-the-art techniques that enable the urban planners to easily describe the issues of interest in free text and, as a result, to accurately identify the emerging concerns from different stakeholders. A first assessment of the London Plan's Green Belt policy has shown the capability of detecting specific community interests about urban planner problems, allowing quick identification of minorities' issues, otherwise overlooked due to the vast amount of data.