What Space Tells Us. Reading spatial traces through a computational-participatory framework for public housing governance

The reality of the spaces we inhabit is shaped by relationships, emotions, and everyday practices that often remain unrepresented—and therefore unconsidered—in governance processes. A partial representation of space risks producing ineffective or unintended interventions. This issue is particularly relevant in public social housing, frequently located in marginalised urban areas, where available data primarily capture vulnerabilities, reinforcing negative narratives and processes of exclusion. Housing, therefore, should not be understood solely as a socio-economic issue of provision, but also as a problem of representation, and consequently a political one. To address this gap, this research proposes a Building Data Management Framework (BDMF) to support public authorities in capturing residents’ lived realities through the observation of how spaces are occupied, adapted, transformed, and maintained. Building on the theory of “errors” developed by Zubčić and Križaj Leko, spatial traces are interpreted not as failures but as expressions of needs, behaviours, emotions, and broader social and cultural dynamics. The framework integrates computational and participatory approaches. Artificial Intelligence is used to process and structure large volumes of spatial data, while the Living Lab methodology provides the human perspective needed to validate and contextualise the results. In particular, Vision-Language Models (VLMs) and Multimodal Machine Learning (MML) support the identification and interpretation of spatial phenomena. The research consists of three stages: a literature review, the development of the BDMF and its simulation through the case study of the Pilastro neighbourhood in Bologna.