Word Wide Web

A word-weaving platform that creates poems through connected synonyms.

2024 / Tool Development, Media Art / Graphic Designer, AI Engineer, Frontend Developer / SFPC Summer 2024

Word Wide Web

Overview

Weaving poems through connected words

Word Wide Web is a word-weaving platform that creates poems through connected synonyms. Inspired by the Korean nursery rhyme ‘Red Like an Apple,’ it helps writers create chain poetry by connecting words through semantic relationships.

Interface for weaving chain poetry

Background

SFPC: Human-Scale Natural Language Processing

During SFPC's Human-Scale Natural Language Processing class, our class collaborated to label words using a system that breaks down text into 16 parts of speech and 6 criteria pairs (like dog-cat, concrete-intangible, utilitarian-romantic). Inspired by this experience, I created a platform that helps writers find similar words based on these semantic measurements.

How it works

Write by connecting words

  1. Enter your starting word.
  2. Browse through suggested related words provided by Word Wide Web.
  3. You can add your own words if you don't find what you're looking for (some words may not exist in our database).
  4. Continue connecting words to craft your poem.

Word suggestion and addition process

Weaving words into poetry

Visit the Word Wide Web website ↗

Final Design

Unexpected connections

Through this process, discover unexpected linguistic connections and create poetic expressions that challenge our perception of language relationships.

Example of unexpected word connections

Posters: word webs and the Verbal Vortex

Credits

Data contribution and tools

Weaving chain poetry with the contributed dataset

Text data, word scoring

Ilona Brand, Lillian Hua, Jean Kim, Maya Williams, kathy wu, Joses Ho, Maya Detwiller, Jireh Deng, Ellen Fritz, Alice Yuan Zhang, Yewon Jang, Charis Poon, Kimberly Lyle, Hannah Jenkins, Cecilia Knaub, Gabriella Chronis, Juliana Gomezr

The data and word scoring were collaboratively created by participants, each developing individual projects based on this shared dataset.

Skills

Python, JavaScript, Label Studio