Sihwa Park

About

Selected Works
YouTube Mirror
Uncertain Facing
GeoD
ARLooper
Ballet Mécanique
Brand Logo Sonification
TopicBubbles
Loading...
BeHAVE
InstaSynth
Structured Improvisation X
FormSound

Archive
SoniPi
Letter Frequency Visualization
Generative Audiovisual Study
Ring of Quartets
Hans Zimmer’s OST vs Film
︎︎︎Past Works (2008-2015)

Recent News

- Oct 24, 2023: Appointed a Connected Minds Research Enhanced Hire position with a $100,000 startup fund.
- Sep 1, 2023: YouTube Mirror was presented at AIMC 2023, Brighton, UK.
- Jun 12, 2023: YouTube Mirror has been accepted for AIMC 2023 as both a paper and an installation!
- May 31, 2023: YouTube Mirror was exhibited at NIME 2023, Mexico City, Mexico.
©2023 Sihwa Park

Sihwa Park



TopicBubbles

An Interactive Topic Model Visualization
Date: Jul 2018 - Jul 2020
Categories: Data Visualization, UI/UX Design

TopicBubbles is a D3-based bubble chart for topic model visualization that aims to enable the user to explore topic models in an intuitive and interactive way. TopicBubbles visualizes topics as circles that are packed at the center of the screen by a force simulation algorithm and that are expandable, movable upon mouse interaction. While the size and color intensity of the circles represent statistical weight in the model, the expansion of circles reveals three more layers of information in detail: a word cloud of top words, top documents, and sources of top documents. Since multiple circles of topics can be open simultaneously on the screen, TopicBubbles stands out in comparison tasks such as comparing multiple topics based on their word clouds or documents, and examining topics including top words of interest. With design elements for closer topic inspection, such as a pie chart visualization for searching for top words of interest, the real-time highlight of top words or documents upon mouse interaction, and an embedded JSON file reader, TopicBubbles helps the user navigate topics in detail.

TopicBubbles was developed as part of the WhatEvery1Says (WE1S) project where I worked as a Research Assistant with the role of Data Visualization Specialist. WE1S uses digital humanities methods to explore public discourse about the humanities in journalistic media available in digital textual format. WE1S works on producing and analyzing topic models from large collections of articles related to the humanities in various media sources in the U.S., the U.K., and Canada from 1981 on. The video above demonstrates one of their topic models created from 82,324 U.S. news articles mentioning “humanities”.

Working example
GitHub repository

Related Project
https://sihwapark.com/GeoD

Related Articles
Alan Liu, Abigail Droge, Scott Kleinman, Lindsay Thomas, Dan C. Baciu, Jeremy Douglass. "What Everyone Says: Public Perceptions of the Humanities in the Media", Daedalus 2022, 151 (3): 19–39. doi: https://doi.org/10.1162/daed_a_01926
Alan Liu, “What Everyone Says about the Humanities”, MLA Newsletter 54, no. 1 (2022): 1, 4–5 [link]