Contemporary urban analysis requires tools and analytical software that can not only capture the physical structure of the city, but also the dynamic and human-centred places that can emerge from everyday interactions. While these technologies, such as existing OpenStreetMap (OSM) [1] views and Geographic Information Systems (GIS) [2] are effective for spatial tasks, these abstractions fail to understand the notions of place when compared to space [3]. Recent work [4, 5] has sought to shift the focus towards platial information systems, tools which situate the human experience, subjective knowledge, and fuzzy representation as a comparison to existing spatial systems. These fuzzy, subjective, and personal representations attempt to model ‘place’ as a contrast to space, which may not align with traditional geometries or formal administrative zones.

PlaceCrafter responds to this challenge by integrating a spatial-platial [6] approach to identifying regions that are functionally cohesive, representing dense and meaningful concentrations of specific points of interest (POI). The POIs are traditional representations of space within GIS [3], such as cafes, restaurants, museums, pathways, and places of worship. Rather than relying on the top-down designation of locations, PlaceCrafter supports users, analysts, and researchers curating clusters that represent how space is used as opposed to administratively divided. The approach aligns with recent calls in GIScience to shift from ‘space’ to ‘place’ in smart city analysis [7] and to continue building on work which has operationalised the sense of place in urban contexts [8].

PlaceCrafter is designed not just to analyse space, but to make its platial structure visible, explorable, and comprehensible to analysts, researchers, and planners. Figure 1 presents the web-based application, developed in TypeScript using React, Vite, Leaflet, Turf.js, and D3.js. The software uses the Overpass API [9] to retrieve (near, depending on number of POIs loaded) real-time user-filtered POI data. The datasets are organised into semantic categories based upon the existing OSM semantic structures [10]; these filters can be customised dependent on the analytical task. This functionality supports the broad purpose of the web-based application, which is to enable researchers and practitioners to understand city form and structure through a platial lens.

PlaceCrafter is structured around four phases guided by the OSM filtering approach: the initial phase (1) focuses on filtering and selection of relevant OSM categories, building upon existing work, such as POI Pulse [11] which classifies regions using semantic signatures, and user behaviour to generate profiles of locations in the Los Angeles area and ClusterRadar [12] which supports comparative spatial clustering and parameter tuning through interactive visualisation to examine how clusters change temporally; the second phase (2) is where the fuzzy clustering approaches are applied interactively and include K-Means [13] for compact cluster formation, DBSCAN [14] for spatial structures, and hierarchical clustering for multi-level spatial structures. These clustering methods are applied to the filtered POI data to reveal platial regions.

The penultimate phase (3) focuses on statistical validation, where each clustered region is evaluated using established spatial metrics. These evaluations include the nearest neighbour index to assess spatial clustering, silhouette scores [15] for understanding cluster coherence, and spatial autocorrelation is measured using a simplified Moran’s I statistic for insights into category-based dependency [16]; The final phase is (4) visualisation, which explores the concept of platial readability, where each region is presented not just spatially but semantically, with data supported by POI type, diversity score, and density metrics. Additionally, the platial visualisation techniques used support the emerging approaches to conveying ambiguity, overlap, and functional gradients [3, 17]. The visualisation subsystem is modular for a wide array of end-user requirements, supporting fuzzy spray can visualisation and region influence grids as presented in Figure 2, in addition to convex hulls, kernel density heatmaps, and region quality indicators.

Figure 3 presents a case study using PlaceCrafter to analyse Nottingham, United Kingdom and the surrounding areas. The POI filtering focused on tourism, historical, leisure, and natural categories. A total of 534 POIs were clustered into 18 functional regions using K-Means. There were 344 historical POIs and 111 leisure POIs as the largest categories from the filtering. The spatial pattern reflects the diverse landscape of Nottingham, with dense clusters in the city centre based around historical POIs, with suburban and rural areas having a more diffuse pattern of historic and leisure clustering. The statistical validation showed strong autocorrelation using Moran’s I (0.68) and a high internal cohesion Silhouette score (0.83), confirming the utility of platial clustering in capturing real-world functional structures.

PlaceCrafter offers a powerful and emerging way to engage with spatial data, but its outputs are shaped by the characteristics of OSM. The crowd-sourced nature of this data means that certain POI types particularly those tied to commerce or tourism, are more visible than the informal or everyday categories more relevant to place-based ambiguity or subjective personal experiences. However, previous research has shown that OSM data is generally reliable for urban-area analysis, despite some noise [18]. Additionally, the clustering algorithms respond to spatial density and POI tag semantics, which do not explicitly capture human-perceived understandings of place boundaries, indicating a need for inclusion of social media data such as those used in POI Pulse [11] or the subjective experiences captured from linked walking narratives in WalkGIS [19].

PlaceCrafter is being developed to be an open-source software which enables broader use, adaptation, and academic contributions in platial information systems. Planned expansions include supporting historical analysis to track the evolution of platial regions and their changes over time. We intend to further expand the platform to incorporate multi-comparative views of platial functional regions, enabling cities, timeframes, and thematic domains to be compared in real-time. We also intend to conduct a qualitative study investigating how users, analysts, and researchers interpret and engage with PlaceCrafters outputs in practice. These improvements will help the validation of the tool’s interpretability and role as a flexible analytical environment for platial analysis.

As cities continue to change, understanding how people engage with their environments becomes increasingly layered and complex. Recognising that cities are more than their administrative boundaries, while valuing the role of spatial data, PlaceCrafter helps uncover the functional geography of place by clustering OSM POIs into meaningful regions that reflect how place is used, shared, and shaped through the spatial logic embedded in mapped data.