Q-Chef: The impact of surprise-eliciting systems on food-related decision-making

Title of the Paper

Q-Chef: The impact of surprise-eliciting systems on food-related decision-making

Paper Information

• Research Domain: Human-Computer Interaction, Recommender System Design, Dietary Behavior Change
• Keywords: Surprise, Curiosity, Food, Recommender Systems, Mixed Methods

Research Background and Problem

• Problems or Challenges:

  1. Traditional recommender systems tend to match user preferences based on similarity, potentially narrowing the range of choices users are exposed to.
  2. There is a direct link between dietary diversity and quality, but many recommender systems fail to effectively promote diversity.
  3. Current research lacks in-depth exploration of how diverse content influences user decision-making.

• Significance: Dietary diversity is closely related to health and nutrition. Increasing the diversity of food choices can not only enhance user experience but also encourage users to try new cooking behaviors, thereby improving dietary quality and health outcomes.

• Research Motivation and Related Work:

  1. Findings from the field of information retrieval suggest that diverse content can increase user satisfaction and engagement.
  2. Surprise and curiosity have been shown in psychology to have the potential to stimulate exploratory behavior.
  3. Designing a recommender system that encourages diverse choices and studying the impact of "surprise" on food-related decision-making is a largely unexplored but highly promising direction.

Solution

• Method or Solution: Propose "Q-Chef"—a prototype application that combines a traditional recipe recommender system with a personalized surprise model, aiming to influence user decisions by offering recipes that align with preferences while incorporating elements of surprise.

• Innovations:

  1. Develop an algorithmic model to predict recipes that users perceive as "surprising," while retaining the "delicious" qualities of these recipes.
  2. Divide the concepts of surprise and curiosity into two factors users might experience: "unfamiliarity" and "unexpectedness," and design a machine learning classifier to predict these aspects.

• Implementation Steps:

  1. Build a user preference model, including familiarity levels and specific taste data.
  2. Use the "surprise model" in combination with preference data to filter recipes.
  3. Classify recommendations based on similarity and degree of surprise, presenting 10 recommended recipes for selection.
  4. Compare user behavior between the "surprise" group and the "deliciousness (control)" group.
  5. Conduct thematic analysis of interview data to uncover deeper motivational factors in user decision-making.

Research Findings

• Specific Findings:

  1. Identified five key drivers of food choices: skill repertoire, comfort and familiarity, personal decision-making mindset, moments of exploration, and social ecology.
  2. Data analysis revealed that users in the "surprise" group were more inclined to recall and explore new experiences, while users in the "deliciousness" group emphasized familiarity and nostalgic connections.

• Advantages:

  1. Compared to traditional systems, it effectively sparks user interest and promotes dietary diversity.
  2. Encourages users to reflect on their food choices, transforming the sense of surprise into exploratory motivation.

• Experimental or Evaluation Results:

  1. Semi-structured interviews and qualitative thematic analysis showed that surprise-based options increased users' acceptance of trying new things.
  2. Trying new recipes led users to associate more with related memories of autonomous exploration and a desire for novelty.

• Limitations and Future Directions:

  1. The study sample was skewed towards younger and highly educated groups, lacking exploration of factors like family with children and cultural differences.
  2. The recipe database was primarily based on Western cultural contexts, with limited adaptability for non-Western cultural groups.
  3. While the surprise model demonstrated high accuracy, personalized predictions require more data support and further optimization.
  4. Future research is recommended to explore how social network-based recommendation mechanisms can support users in trying diverse content, while further investigating how to optimize the long-term effects of user engagement.

Conclusion

The study provides cutting-edge insights into the design of surprise-based recommender systems and their impact on user experience, offering novel recommendations for designing systems that stimulate curiosity. It demonstrates significant potential, particularly in promoting healthy dietary behaviors.