Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Decade of Systems for Human Data Interaction

Eugene Wu, Yiru Chen, Haneen Mohammed, Zezhou Huang

TL;DR

This work argues that human–data interaction (HDI) requires tight co-design of interfaces and data systems, coining Design Dependence to describe mutual constraints that shape interactivity, latency, and perceived performance. It proposes principled abstractions—Physical Visualization Design (PVD) and its Jade instantiation, and the Data Interface Grammar (DIG)—to synthesize end-to-end HDI architectures and automate optimization across heterogeneous backends. It further shows that system-level abstractions can inspire new interfaces (View Composition Algebra, multi-table analytics, database visualization) and that database theory enables semantically faithful multi-table visualizations, expanding the space of interactive analytics. The paper concludes that AI will augment but not replace HDI systems, underscoring the need for robust abstractions, provenance, and performance guarantees to support interactive, trustworthy AI-driven data reasoning.

Abstract

Human-data interaction (HDI) presents fundamentally different challenges from traditional data management. HDI systems must meet latency, correctness, and consistency needs that stem from usability rather than query semantics; failing to meet these expectations breaks the user experience. Moreover, interfaces and systems are tightly coupled; neither can easily be optimized in isolation, and effective solutions demand their co-design. This dependence also presents a research opportunity: rather than adapt systems to interface demands, systems innovations and database theory can also inspire new interaction and visualization designs. We survey a decade of our lab's work that embraces this coupling and argue that HDI systems are the foundation for reliable, interactive, AI-driven applications.

A Decade of Systems for Human Data Interaction

TL;DR

This work argues that human–data interaction (HDI) requires tight co-design of interfaces and data systems, coining Design Dependence to describe mutual constraints that shape interactivity, latency, and perceived performance. It proposes principled abstractions—Physical Visualization Design (PVD) and its Jade instantiation, and the Data Interface Grammar (DIG)—to synthesize end-to-end HDI architectures and automate optimization across heterogeneous backends. It further shows that system-level abstractions can inspire new interfaces (View Composition Algebra, multi-table analytics, database visualization) and that database theory enables semantically faithful multi-table visualizations, expanding the space of interactive analytics. The paper concludes that AI will augment but not replace HDI systems, underscoring the need for robust abstractions, provenance, and performance guarantees to support interactive, trustworthy AI-driven data reasoning.

Abstract

Human-data interaction (HDI) presents fundamentally different challenges from traditional data management. HDI systems must meet latency, correctness, and consistency needs that stem from usability rather than query semantics; failing to meet these expectations breaks the user experience. Moreover, interfaces and systems are tightly coupled; neither can easily be optimized in isolation, and effective solutions demand their co-design. This dependence also presents a research opportunity: rather than adapt systems to interface demands, systems innovations and database theory can also inspire new interaction and visualization designs. We survey a decade of our lab's work that embraces this coupling and argue that HDI systems are the foundation for reliable, interactive, AI-driven applications.

Paper Structure

This paper contains 21 sections, 11 figures.

Table of Contents

  1. Introduction
  2. From Interface to System Design
  3. Black-box Query Workload
  4. Interface Properties
  5. Network Communication
  6. Approximation
  7. A Deluge of Special Cases
  8. Formal Representations & Design Independence
  9. Physical Visualization Design
  10. An Instance of Physical Visualization Design
  11. What Next?
  12. From Systems to New Interfaces
  13. From Query Workloads to New Interactions
  14. From Query Workloads to New Interfaces
  15. From Single Table to Multi-Table Analytics
  16. ...and 6 more sections

Figures (11)

  • Figure 1: Design dependence describes how interface designs necessitate system complexity, but system designs also limit the capabilities of the interface. This dependence necessitates a holistic approach to human data interface design. The dependencies are labeled with the sections that describe them.
  • Figure 2: Interface designs necessitate new system designs to meet analysis needs and interactivity expectations. Systems are increasingly shifting from a black box model of the interface, to one that leverages specific interface properties, to one that leverages a formal representation of the interface.
  • Figure 3: (a) Interface to analyze vote counts by congressional member. (b) Interactions bind to choices in a logical query plan. Three candidate execution plans that (c) send queries to the cloud DBMS, (d) pre-compute and evaluate interactions using cubes on the server, and (e) caches cubes on the client.
  • Figure 4: Pareto curve for the client and server resources needed to meet latency expectations for 6 interactive interfaces.
  • Figure 5: Database capabilities and theory go well beyond the system designs and optimizations developed in response to today's interface designs, and can motivate new types of interface and interaction designs, as part of a virtuous cycle.
  • ...and 6 more figures

Theorems & Definitions (5)

  • Example 1
  • Example 2
  • Example 3
  • Example 4
  • Example 5