Designing Persuasive Food Conversational Recommender Systems With Nudging and Socially-Aware Conversational...

Area of Science:

• Human-Computer Interaction
• Artificial Intelligence
• Public Health

Background:

• Unhealthy eating behaviors pose significant public health challenges.
• Conversational systems offer a potential solution for promoting healthier dietary choices.
• Personalization is key for effective recipe recommendations.

Purpose of the Study:

• To develop and evaluate Cora, a conversational system for personalized healthy recipe recommendations.
• To assess the impact of conversational skills and interaction modes on user perception.
• To investigate the influence of explanations and comparative recommendations on user engagement.

Main Methods:

• Cora, a conversational system, was developed to recommend recipes based on user habits and preferences.
• Two experiments were conducted: one on conversational skills and interaction modes, the other on explanations and comparisons.
• User perception, intention to cook, and system usefulness were measured.

Main Results:

• Rapport-building dialogue significantly improved user perception of the interaction and the system.
• Explanations positively influenced user perception of the recommender system.
• Comparing healthy recipes with a decoy increased perceived usefulness, but detailed explanations reduced system adoption likelihood.

Conclusions:

• Conversational AI can effectively promote healthy eating through personalized recipe recommendations.
• Social dialogue and clear explanations enhance user experience with recommender systems.
• Careful design is needed when using comparative recommendations to avoid unintended negative impacts on user adoption.