DiTTo: The Taxonomy

• Timeliness — From stream-processed sub-second updates to weekly/monthly summaries; includes event framing for decision-making.
• Fidelity — From abstract lists/spreadsheets to high-resolution 2D/3D/VR scenes; choose fidelity to match task and data suitability.
• Aggregation — From per-sensor/per-person detail to privacy-preserving groupings across spaces or systems.

The axes can be paired (T/F, F/A, T/A) to compare designs, spot gaps, and plan interfaces. For instance, highly aggregated views often trend toward lower timeliness, suggesting opportunities for privacy-preserving real-time designs.

Axes

The taxonomy’s three axes — timeliness, fidelity, and aggregation — are best explored interactively through the demo above. You can view each pairing (T/F, F/A, T/A) using the buttons to compare designs and spot gaps. For static reference diagrams and full details, see the chapter PDF.

Case Study Summary

The LT1 testbed integrates environmental sensors (CO₂, temperature, humidity) and human-centric data (crowd count/localisation), streaming 1–4× per minute at the sensor level. Visualisations bind data to space (2D/3D floorplans) and time (streams and history), using four atomics (floorplan, heatmap, basic charts, lists). Plotted against the three axes, the set reveals missing corners (e.g., privacy-preserving real-time, or low-timeliness/low-aggregation exemplars) which the taxonomy helps uncover.

The Effects of Visualisation Fidelity

A key question in my thesis was how visualisation fidelity affects user interpretation and decision-making, particularly with respect to timeliness. Lower-fidelity visualisations (e.g. lists, 2D heatmaps) are often faster to interpret and act upon, making them well suited for real-time monitoring. Higher-fidelity visualisations (e.g. 3D models, realistic renderings) can be more engaging and informative for analytical tasks, but may introduce cognitive overhead and slow down decision-making in time-critical contexts. In short, fidelity should be matched to the task and data, rather than defaulting to high fidelity for its own sake.

From Visualisations to Interfaces

Real products compose multiple DT representations into tools (dashboards, rewind/scrub, anomaly views). The taxonomy guides coverage across quadrants so interfaces balance monitoring, analysis, and maintenance, while respecting privacy.

Conclusions

DiTTo is deliberately compact for practicality. It provides a shared language for positioning designs, reasoning about trade-offs, and clarifying that ‘real-time’ refers to stream processing and event sense-making rather than simple recency. Above all, it centres user agency by enabling designers to compare visualisations and identify novel solutions.