Skip to main content

Annotation Design Space & Taxonomy

Read Paper
A Qualitative Analysis of Common Practices in Annotations: A Taxonomy and Design Space
Md Dilshadur Rahman, Ghulam Jilani Quadri, Bhavana Doppalapudi, Danielle Albers Szafir, Paul Rosen
IEEE Transactions on Visualization and Computer Graphics 2024
Overview Design Space Methodology Taxonomy Case Studies
Supplemental ↗
01 · Overview

Contributions

A qualitative study of 1,888 annotated charts across 14 chart types. The paper identifies seven annotation types and organizes annotation design around task, annotation form, and data source.

Key Contributions

  • Taxonomy: We analyze N=1,888 annotated charts across 14 chart types and identify seven recurring annotation types. The taxonomy is based on observed practice in real charts.
  • Design Space: We organize annotation decisions with three questions: [Why] for the task, [How] for annotation types and ensembles, and [What] for whether annotation content is internal, derived, or external.
  • Corpus & Case Studies: We provide the annotated chart dataset and use three published charts to show how the design space can guide annotation choices.
02 · Design Space

Why, How, and What

The paper frames annotation design with three questions. Why asks what task the annotation supports. How covers the annotation type or ensemble used to support that task. What identifies whether the annotation content comes from the chart data, from a calculation, or from outside knowledge.

Original design-space figure from the paper showing Why, How, and What relationships across annotation types
Original figure from the paper. The left panel shows task support. The right panel shows annotation usage by chart type and ensemble. The bottom panel groups annotations by data source.

Design Space Guide

Why: Task intent. Begin with the task. All seven annotation types support Identify. Most also support Compare. Text, glyphs, and indicators support Summarize. Only text supports Present.
How: Form and ensemble. Select the annotation type that fits the task. The paper identifies seven types and shows that many charts use ensembles when one type is not enough. Connector plus text is a common example.
What: Data provenance. Content source can be internal, derived, or external. The same annotation type can draw from different sources depending on the message the author wants to communicate.
03 · Methodology

Dataset Construction

The dataset came from Google Images and was narrowed through duplicate removal and chart screening. Two coders then iteratively labeled the remaining charts to identify recurring annotation types and usage patterns.

8,768
Retrieved images
2,677
Candidate charts
1,888
Charts analyzed
14
Chart types
7
Annotation types
0.886
Average Cohen's kappa
Study Figure
Cropped methodology figure from the paper showing collection filtering and coding workflow
After duplicate removal and chart screening, 2,677 candidate charts remained. The final analysis used 1,888 charts from one coder's labels across 14 chart types.
Study Process
  1. Collection Google Images

    Retrieved 8,768 annotated chart images across 14 chart types.

  2. Screening 2,677 candidates

    Removed 1,244 duplicates and excluded 4,847 charts that were out of scope, unclear, unannotated, or not based on real data.

  3. Coding 7 final types

    Two coders worked in multiple rounds, starting from five annotation types from prior work and refining them into seven recurring types.

Reliability: The first coder retained 1,888 charts for analysis. The average Cohen's kappa across coders was 0.886.

04 · Taxonomy

Seven Annotation Types

These seven cards summarize the annotation types reported in the paper. Open a card to see the tasks each type supports, common ensembles, likely data sources, and where it appears in the case studies.

Filter and Explore

Filter by task. The same initials shown on the filters are repeated on each annotation card.

05 · Case Studies

Case Studies

The paper applies the design space to three published charts. Each case shows how annotation choices support a task, take a visual form, and draw from a data source.

How to Read

Select a case. The left side shows the annotated figure and main points. The right side lists the annotation rows by task, form, and source.

06 · Citation

Cite This Work

@article{rahman2024annotation,
  title   = {A Qualitative Analysis of Common Practices in
             Annotations: A Taxonomy and Design Space},
  author  = {Rahman, Md Dilshadur and Quadri, Ghulam Jilani
             and Doppalapudi, Bhavana and Szafir, Danielle Albers
             and Rosen, Paul},
  journal = {IEEE Transactions on Visualization and Computer Graphics},
  year    = {2024},
  doi     = {10.1109/TVCG.2025.3565855}
}