Document Type

Poster

Publication Date

Fall 2024

Abstract

This research presents a topic modeling analysis of the WILDCHAT-FULL dataset, examining user interactions with ChatGPT across over one million conversations. The study focused specifically on extended conversations (five or more exchanges) between U.S.-based users and ChatGPT in English. Using Latent Dirichlet Allocation (LDA), I identified 50 distinct conversational topics with coherence scores ranging from -1.5 to -14.92 (mean: -4.47). The analysis revealed diverse interaction patterns spanning creative writing, jailbreaking (attempting to get ChatGPT to do something against its guidelines), technical discussions, business applications, and educational queries. A particularly striking finding was that creative writing and role-play scenarios dominated the interactions, comprising over 25% of the identified topics.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.