Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

User Experience with LLM-powered Conversational Recommendation Systems: A Case of Music Recommendation

Sojeong Yun, Youn-kyung Lim

TL;DR

This paper investigates how LLM-powered CRS enable user-driven, customized music recommendations beyond traditional RS. Through a three-week diary study with 12 participants using personalized CRS-GPTs, it shows that such systems help users articulate implicit needs, explore music in novel ways, and gain deeper understanding of their own preferences. The authors identify a design space for CRS that emphasizes self-discovery, designability, and deliberate use of ambiguity to support critical reflection. These insights suggest practical implications for building more personalized, user-controlled recommendation experiences that empower users as active service designers rather than passive recipients.

Abstract

The advancement of large language models (LLMs) now allows users to actively interact with conversational recommendation systems (CRS) and build their own personalized recommendation services tailored to their unique needs and goals. This experience offers users a significantly higher level of controllability compared to traditional RS, enabling an entirely new dimension of recommendation experiences. Building on this context, this study explored the unique experiences that LLM-powered CRS can provide compared to traditional RS. Through a three-week diary study with 12 participants using custom GPTs for music recommendations, we found that LLM-powered CRS can (1) help users clarify implicit needs, (2) support unique exploration, and (3) facilitate a deeper understanding of musical preferences. Based on these findings, we discuss the new design space enabled by LLM-powered CRS and highlight its potential to support more personalized, user-driven recommendation experiences.

User Experience with LLM-powered Conversational Recommendation Systems: A Case of Music Recommendation

TL;DR

This paper investigates how LLM-powered CRS enable user-driven, customized music recommendations beyond traditional RS. Through a three-week diary study with 12 participants using personalized CRS-GPTs, it shows that such systems help users articulate implicit needs, explore music in novel ways, and gain deeper understanding of their own preferences. The authors identify a design space for CRS that emphasizes self-discovery, designability, and deliberate use of ambiguity to support critical reflection. These insights suggest practical implications for building more personalized, user-controlled recommendation experiences that empower users as active service designers rather than passive recipients.

Abstract

The advancement of large language models (LLMs) now allows users to actively interact with conversational recommendation systems (CRS) and build their own personalized recommendation services tailored to their unique needs and goals. This experience offers users a significantly higher level of controllability compared to traditional RS, enabling an entirely new dimension of recommendation experiences. Building on this context, this study explored the unique experiences that LLM-powered CRS can provide compared to traditional RS. Through a three-week diary study with 12 participants using custom GPTs for music recommendations, we found that LLM-powered CRS can (1) help users clarify implicit needs, (2) support unique exploration, and (3) facilitate a deeper understanding of musical preferences. Based on these findings, we discuss the new design space enabled by LLM-powered CRS and highlight its potential to support more personalized, user-driven recommendation experiences.

Paper Structure

This paper contains 25 sections, 7 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Supporting active participation of users in CRS
  4. LLM-powered CRS
  5. Method
  6. Recommendation System Domain
  7. Participants
  8. Diary Study Design
  9. Week 1: Exploration Week – Exploring Various Ideas Across the Three Stages of Recommendation Process
  10. Week 2: Selecting Week – Experiencing and Evaluating the First Customized Recommendation System
  11. Week 3: Finalization Week – Experiencing and Evaluating the Refined Customized Recommendation System
  12. Interview Procedure
  13. Overall Results and Data Analysis
  14. Overall Results
  15. Data Analysis
  16. ...and 10 more sections

Figures (7)

  • Figure 1: Three-week study schedule
  • Figure 2: P5’s completed icebreaking sheet, (a) evaluating each scenario based on the researcher-provided tasks, (b) brainstorming personal ideas based on prior experiences
  • Figure 3: P6’s activity using ChatGPT for music recommendation
  • Figure 4: P7’s planning sheet and the GPT customization process (blue text indicates participant input)
  • Figure 5: P3’s screenshot using the custom GPT to receive pop music recommendations based on preferences for K-pop and J-pop
  • ...and 2 more figures