用户体验视角下 ChatGPT 辅助信息检索的可用性研究

Translated title of the contribution: Usability Study of ChatGPT-Assisted Information Retrieval from the Perspective of User Experience

Lei Li, Zhao Liu, Xuyan Wang

Research output: Contribution to journalArticlepeer-review

Abstract

[Purpose/ significance] This study explores the usability of ChatGPT-assisted information retrieval from a user experience perspective, aiming to provide valuable insights for the application of ChatGPT and similar technologies in the field of information retrieval. [Method/ process] Combining experimental and interview methods, an experimental environment was set up using the ChatGPT for Google search assistant plugin alongside a search engine. Retrieval tasks were designed, and 10 participants were asked to complete the tasks with the assistance of ChatGPT. Following this, interviews were conducted to gather participants’ experiences and feelings during the ChatGPT-assisted retrieval. The interview data were then analyzed qualitatively using iterative thematic coding, leading to conclusions about the usability of ChatGPT-assisted information retrieval. [Result/ conclusion] The analysis across five thematic dimensions-user experience, information quality, assistance effectiveness, human-computer interaction, and areas for improvement-indicated that the ChatGPT-assisted retrieval model possesses a certain degree of usability. In this proposed assistance model, ChatGPT plays an intuitive supporting role, significantly reducing search time and providing a satisfactory overall experience, particularly in creative tasks where the assistance effect is more pronounced. ChatGPT’s advantage in information integration and presentation allows it to quickly provide users with key points. However, it was found that ChatGPT may be inaccurate or lack depth when answering specific professional or data-intensive questions, suggesting that traditional search engines could be used to verify and complement ChatGPT’s output. The study confirms the potential of ChatGPT-assisted information retrieval in helping users efficiently and effectively obtain high-quality information.

Translated title of the contributionUsability Study of ChatGPT-Assisted Information Retrieval from the Perspective of User Experience
Original languageChinese (Traditional)
Pages (from-to)75-82
Number of pages8
JournalInformation studies: Theory and Application
Volume48
Issue number2
DOIs
StatePublished - Feb 2025
Externally publishedYes

Fingerprint

Dive into the research topics of 'Usability Study of ChatGPT-Assisted Information Retrieval from the Perspective of User Experience'. Together they form a unique fingerprint.

Cite this