Bestselling Books Project
- Data is art finalist UVA School of Data Science
- Information is Beatiful Award 2024 Longlist
- Medium
- Interactive data visualization and video
- Materials
- Data
- Dimensions
- Web; video, 1920 × 1080 px
- Tools
- p5.brush, dataset, LLM (data cleaning and coding)
The Story
The Bestselling Books Project is an artistic data visualization that treats literary genre not as a fixed taxonomy but as a fluid social construct. Taking the Barthesian premise that to read is to struggle to name, the Bestselling Books Project maps the networks of community-generated subject labels that readers actually use to describe bestselling books. By analyzing bestseller books data and user-generated book labels from Open Library's API, the project renders the relationships among the 200 most-used subject tags and the books they cluster, inviting viewers to explore genre boundaries that prove more contested than fixed.
The Bestselling Books Project is organized as an exploratory interface that reveals unexpected connections between genres while preserving how readers actually describe and connect with books. The project’s interface expresses this network from two directions: a Network View, where node size encodes a subject's total connections, and a Genre Tag Library, where each book is sized by sales and brushstroke color and style encode its original language of publication.
The interface was built using JavaScript, p5.js brush and LLM collapsing approximately 80 raw labels to 11 navigable types. Our interactive network maps the relationships between subjects and their most-sold books, filtered to include the top 20 bestsellers per subject. The visualization combines quantitative network analysis with artistic data visualization research to transform complex bibliographic data into an intuitive, exploratory experience. While our dataset draws from open-source data, capturing neither every bestseller ever written nor sales details, this limitation aligns with our observation: literary classification, like any categorization system, is incomplete and ever-evolving. We invite users to explore these connections and contribute your own observations to this ongoing dialogue.
Credits
Scientific collaborator: Yixuan Liu
Music: Jose Gonzalez