Investigating the Potential of Group Recommendation Systems As a Medium of Social Interactions: A Case of Spotify Blend Experiences between Two Users

Title of the Paper

Exploring the Potential of Group Recommender Systems as Social Interaction Media: A Case Study of Spotify Blend

Bibliographic Information

• Subject Area: Human-Computer Interaction (HCI), User Experience Design, Role of Recommender Systems in Social Contexts
• Keywords: Group Recommender Systems (GRS), Social Interaction through Technological Media, User Experience, Social Dynamics, Spotify Blend, HCI, Recommendation Algorithms, Social Technology, Music Sharing, Social Behavior

Research Background and Issues

• Problems or Challenges Identified by the Authors:

  1. Previous research on Group Recommender Systems (GRS) has focused more on recommendation accuracy and efficiency, with limited exploration of how social interaction impacts user experience.
  2. The potential of GRS not only to provide personalized recommendations but also to serve as a medium for social interaction remains underexplored.
  3. Existing technological designs often fail to adequately capture the implicit and complex characteristics of users' social behaviors.

• Significance of the Problem:

  • Understanding the role of social dynamics and interaction in the context of GRS is crucial for optimizing the user experience in such systems.
  • GRS has the potential to facilitate indirect communication between users and potentially strengthen interpersonal relationships, opening up possibilities for its role as an emerging social technology.

• Research Motivation and Related Work:

  1. Using Spotify Blend as a real-world example of a typical GRS, the study explores its role in complex social contexts.
  2. Prior research suggests that while the primary function of GRS is to support group decision-making, it also has the potential to enhance social emotions and facilitate everyday communication between users.
  3. Existing literature (e.g., shared playlists and media recommender systems) provides insights into design strategies tailored for specific user groups but has yet to systematically explore the social attributes of GRS.

Proposed Solution

• Proposed Methods/Solutions:

  • Based on the Spotify Blend service, the study employs a mixed-methods approach combining qualitative research and longitudinal tracking to investigate social interactions between two users within a GRS.
  • The research design includes a preliminary survey (online questionnaire), followed by a 21-day diary study and in-depth semi-structured interviews during the main study phase.

• Innovative Contributions:

  • The study introduces the hypothesis of GRS as a "social catalyst," suggesting that it not only provides personalized recommendations but also indirectly promotes social exchange between users.
  • It proposes two core attributes for designing GRS to optimize social experiences: the ambiguity of recommendation algorithms and the visibility of user behaviors.

• Implementation Steps and Key Techniques:

  1. Preliminary Survey: Gathered Spotify Blend usage experiences from 38 users to narrow the research focus on social interaction in dyadic relationships.
  2. 21-Day Diary Study: Collected daily usage experiences and behavioral data from 30 participants (15 pairs with varying interpersonal relationships).
  3. In-Depth Interviews: Conducted individual interviews to refine participants' diary feedback and gain deeper insights into nuanced social contexts.
  4. Data Analysis: Employed thematic analysis to identify key themes related to users' social behaviors (e.g., implicit social interactions, indirect communication methods).

Research Findings

• Specific Findings:

  1. Spotify Blend highlights the potential of GRS in facilitating indirect social interactions. Users exhibited implicit behaviors during usage, including:
     • Developing mutual understanding of each other's music preferences, emotional states, and contexts.
     • Using recommendations to make indirect, non-intrusive suggestions (e.g., influencing shared recommendations through listening choices).
     • Demonstrating prosocial behaviors (e.g., considering a partner's preferences and fostering shared interests).
  2. User experience was significantly influenced by the ambiguity of recommendation algorithms and the visibility of user behaviors. These design elements indirectly encouraged deeper inference and understanding of one another.

• Comparison with Existing Solutions and Advantages:

  • Compared to existing studies that focus more on recommendation accuracy, this research provides a novel perspective on the social dimensions of GRS.
  • By analyzing users' everyday interactions, the study reveals the value of implicit social connections, offering concrete guidance for future GRS design.

• Experimental or Evaluation Results:

  • Findings indicate that users often use GRS to facilitate unspoken communication, thereby enhancing intimacy and resonance with one another.
  • The proposed design attributes (ambiguity and visibility) are identified as potential keys to creating more nuanced social interaction experiences.

• Limitations and Future Directions:

  1. Sample Limitations: The study is limited to South Korean users and focuses on dyadic interactions; further research is needed to explore multi-user scenarios and cross-cultural perspectives.
  2. Domain Limitations: The study only examines the music recommendation domain; other domains, such as video or news recommendation, may exhibit different characteristics.
  3. Long-Term Effects: Some deeper relational changes may not have fully manifested within the 21-day timeframe, requiring long-term studies for validation.
  4. Analysis of Specific Relationships: Future research should further investigate how different types of interpersonal relationships (e.g., friends vs. colleagues) influence GRS experiences.