Buying the 'Right' Thing: Designing Food Recommender Systems with Critical Consumers

Document Title

Buying the ‘Right’ Thing: Designing Food Recommender Systems with Critical Consumers

Document Information

• Subject Area: Recommender Systems, Consumer Ethics, Food Choices
• Keywords: Critical Consumerism, Recommender Systems, Consumer Informatics, Food Ethics, Co-Design, Knowledge-Driven

Research Background and Issues

• Issues and Challenges: Critical consumers face difficulties in balancing their ethical values, budget constraints, and practical needs in complex consumption decisions. Current recommender systems are often designed from the perspective of merchants, lacking transparency and personalization, and are unable to effectively support niche practices (e.g., veganism or plastic-free consumption).
• Significance: As sustainability and social responsibility issues gain increasing attention, designing systems that help consumers make choices aligned with their values can drive change in market-driven societies.
• Research Motivation: Existing academic and commercial recommender systems overlook the diversity of niche needs and ethical values, and there is limited research on how recommender systems can effectively support ethical consumption.

Solution

• Method and Design: This study proposes a knowledge-based food recommender system (F-RS4CC) that collects personalized consumer data (e.g., allergy restrictions, ethical goals, and budget) and translates it into specific product recommendations.
• Innovations:
  • The system integrates consumers' ethical values, dietary restrictions, and availability constraints, allowing users to flexibly adjust weights and priorities.
  • Incorporates "place" to optimize purchases by considering retail infrastructure, addressing the lack of transparency in current recommender systems.
  • Utilizes multi-source data (e.g., retailer information and the open-source data platform Open Food Facts) to enrich product information.
• Implementation Steps:
  1. Pre-study: Conducted interviews with 24 critical consumers to understand existing consumption practices and pain points.
  2. Co-Design: Explored design solutions with consumers in a participatory workshop.
  3. Development and Implementation: Developed a prototype system based on survey and discussion results, with features including "exploring new products" and "generating shopping lists."
  4. Evaluation: Assessed the system through a two-month usage period by 10 consumers, followed by subsequent interviews.

Research Outcomes

• Specific Outcomes:
  • Designed a prototype personalized food recommender system supporting critical consumers.
  • The study demonstrated that the system significantly reduced the complexity of exploring new products and stores while enhancing users' self-reflection capabilities.
  • The system assisted consumers in quickly analyzing suitable products and infrastructures in new environments (e.g., relocating to a new city).
• Comparative Advantages:
  • Compared to traditional recommender systems, F-RS4CC more comprehensively considers ethical values and personal constraints.
  • The system prioritizes stores and products based on consumers' actual circumstances (e.g., budget or distance), improving convenience and personalization.
• Experimental and Evaluation Results:
  • Found that the recommender system primarily supports product exploration rather than shopping planning.
  • Some consumers adjusted their value settings (e.g., carbon footprint) and stopped purchasing certain products through the system.
  • The evaluation highlighted the critical role of data reliability (e.g., mislabeling and tagging) in building consumer trust and suggested supporting more advanced customization features.
• Limitations and Future Directions:
  • Data quality and algorithm transparency need improvement, as the system heavily relies on the accuracy of data providers.
  • Community sharing and rule development features warrant further exploration.
  • Future research could focus on groups without established consumption patterns and the system's impact on long-term consumption behavior changes.

Through this study, the authors demonstrate the feasibility of using knowledge-driven recommender systems to help critical consumers practice their ethical values, while revealing how consumers use technology to balance complex consumption needs and constraints. This provides valuable insights into proposing personalized technologies that support sustainable consumption.