# Broadly defining the term 'Data Visualisation'

Canonical URL: https://jeanyoon.ch/oi/texts/broadly-defining-the-term-data-visualisation
Type: essay / research note
Author: Jeanyoon Choi (최정윤)
Date: 2025-02-09

## Summary

This 2025 note proposes a broad definition of data visualisation for computational art: text fields, interactive 3D forms, images, colours and lighting computed from inputs are all visualisations of processed data, as distinct from fixed playback such as a prerecorded film. Jeanyoon Choi argues that as works become modular, their identity lies in inputs, connections and algorithms rather than prepared assets. Generative AI extends this further, since language models can generate not only material but the connections themselves, pointing toward modular, automatically operating platforms.

## 요약 (Korean)

최정윤은 계산 예술에서 '데이터 시각화'를 그래프나 대시보드를 넘어, 입력·연결·알고리즘으로 출력이 결정되는 모든 웹·소프트웨어 작업으로 넓게 정의한다.

## Questions this answers

### Is generative or interactive web art a form of data visualisation?

Jeanyoon Choi argues yes, in a broad sense: in software art, data is gathered, transformed through relationships and presented, so text, 3D forms, images and lighting computed from inputs are visualisations of data. A prerecorded film, playing a fixed sequence of pixels, falls outside this definition.

### How does generative AI change data visualisation in art?

In 'Broadly defining the term Data Visualisation', Choi notes that generative AI lets small inputs produce large outputs and lets language models generate algorithms, so not only assets but the connections between them can be generated.

Topics: data visualisation, computational art, software art, generative ai, modular systems

## Full text

Data visualisation usually brings graphs and dashboards to mind. I want to use the term more broadly within computational art.

In software art, data is gathered, transformed through relationships and presented. A field of text is processed data made visible. A three-dimensional form that changes through interaction is also a visualisation of data. So are images, colours and lighting whose states are calculated from inputs.

There is a point at which that definition becomes unhelpfully broad. A prerecorded film also consists of data, but playing a fixed sequence of pixels is different from data actively determining a work’s changing output. For this discussion, I am concentrating on web and software artworks whose parameters remain operative.

The more modular and computational the work becomes, the less it depends on a fixed collection of prepared assets. Its identity increasingly lies in the inputs, connections and algorithms through which outputs emerge. In that sense, what it presents is a visualisation of processed relationships.

Generative AI expands the range of those transformations. A small amount of text can produce a much larger image; the text itself could come directly from a participant or be derived from a nonlinguistic action. A language model can also generate an algorithm, subject to whatever validation is needed before it is used.

That changes the proposition. Previously, I might generate an asset while keeping the connections fixed. Now the connections themselves could be generated. The work could become a modular, automatically operating platform in which both material and relationships vary.

What would make this concrete? What example would communicate the difference clearly? Those remain the practical questions behind this expanded definition.

---

Jeanyoon Choi (최정윤) is a Korean computational artist and researcher (b. 1999), PhD candidate at KAIST XD Lab, making multi-device web artworks. Profile: https://jeanyoon.ch/oi/en. Index: https://jeanyoon.ch/llms.txt.