Curio: A Dataflow-Based Framework for Collaborative Urban Visual Analytics

TL;DR

Curio addresses the fragmentation of urban analytics tools by introducing a provenance-aware dataflow framework that supports code, grammar, and GUI abstractions. It enables asynchronous collaboration among urban experts, visualization researchers, and practitioners through a shared canvas and node-level templates. Key contributions include a formal Curio dataflow model, a library of urban-specific modules (loading, transformation, analysis, visualization, interaction), and a provenance mechanism. The evaluation via usage scenarios in urban accessibility, microclimate, and shadow/visualization of heterogeneous data demonstrates Curio's flexibility and potential impact on reproducibility and cross-domain collaboration.

Abstract

Over the past decade, several urban visual analytics systems and tools have been proposed to tackle a host of challenges faced by cities, in areas as diverse as transportation, weather, and real estate. Many of these tools have been designed through collaborations with urban experts, aiming to distill intricate urban analysis workflows into interactive visualizations and interfaces. However, the design, implementation, and practical use of these tools still rely on siloed approaches, resulting in bespoke applications that are difficult to reproduce and extend. At the design level, these tools undervalue rich data workflows from urban experts, typically treating them only as data providers and evaluators. At the implementation level, they lack interoperability with other technical frameworks. At the practical use level, they tend to be narrowly focused on specific fields, inadvertently creating barriers to cross-domain collaboration. To address these gaps, we present Curio, a framework for collaborative urban visual analytics. Curio uses a dataflow model with multiple abstraction levels (code, grammar, GUI elements) to facilitate collaboration across the design and implementation of visual analytics components. The framework allows experts to intertwine data preprocessing, management, and visualization stages while tracking the provenance of code and visualizations. In collaboration with urban experts, we evaluate Curio through a diverse set of usage scenarios targeting urban accessibility, urban microclimate, and sunlight access. These scenarios use different types of data and domain methodologies to illustrate Curio's flexibility in tackling pressing societal challenges. Curio is available at https://urbantk.org/curio.

Figures (5)

• Figure 1: Illustration of the key concepts of the Curio dataflow model. (a) A thematic layer $l_c$ and physical layers $l_b$ and $l_n$ are loaded. (b) Spatial joins between $l_c$ and $l_b$ ($l_c\bigotimes l_b$), and $l_c$ and $l_n$ ($l_c\bigotimes l_n$) are computed. (c) The results of the joins are visualized. (d) To support linked views, the Curio dataflow makes use of interaction nodes. (e) $l_c\bigotimes l_b$ and $l_c\bigotimes l_n$ are further joined, creating a link between them. (f) Interaction nodes augment the previously joined layers, propagating information when a user selects or brushes elements in a visualization (shown in (g)).
• Figure 2: Left: Main elements of Curio's interface. Center, Right: Connecting facets from the same node. Center: A drop-down menu listing mark attributes is created through an annotation in a Vega-Lite specification. Right: A checkbox is created, but now through an annotation in Python code.
• Figure 3: Curio's frontend and backend components.
• Figure 4: Using Curio to facilitate expert-in-the-loop inspection of a computer vision model. (a) The user begins by training the model. Provenance information is stored, allowing them to revert to previous versions of the model or explore different training parameters. (b) New nodes are created to load unseen image data and compute the uncertainty of predictions. (c) A physical layer describing neighborhoods in Boston is loaded. (d) An interactive visualization is created, enabling experts to analyze prediction uncertainty across neighborhoods in Boston.
• Figure 5: Using Curio to create visualizations leveraging multiple datasets. (a) The user loads weather data and computes the UTCI, (b) followed by a spatial join with a physical layer describing neighborhoods in Milan. (c,d) The user then creates linked visualizations that highlight neighborhoods with high UTCI and a large population of older adults. (e) By changing two nodes, the user can create a similar visualization using data from Chicago.