2026-05-16T20:17:13Zhttps://keep.lib.asu.edu/oai/request

oai:keep.lib.asu.edu:node-2012802025-05-23T22:12:45Zoai_pmh:alloai_pmh:repo_items

201280 https://hdl.handle.net/2286/R.2.N.201280 http://rightsstatements.org/vocab/InC/1.0/ http://creativecommons.org/licenses/by-nc-sa/4.0 2025-05 63 pages Lin, Waley Echeagaray, Maria Ortiz, Michael Barrett, The Honors College Computer Science and Engineering Program As interest in food and beverage education grows among consumers and professionals alike, there is increasing demand for learning tools that are accessible, interactive, and tailored to individual needs. Traditional methods such as static tasting guides or instructor-led classes often lack personalization and real-time feedback, limiting their impact on learners with varying experience levels. In response to this gap, this thesis presents the design, development, and evaluation of the Sip & Savor Study chatbot: an AI-powered virtual sommelier embedded within a mobile application that supports personalized beverage education. Built using React Native, FastAPI, Firebase, Render and OpenAI’s GPT-4 API, the chatbot delivers real-time recommendations and interactive learning experiences across four major beverage categories: wine, beer, saké, and cocktails. It dynamically adapts to user preferences (e.g., dietary needs, favorite drink types, and taste profiles) and supports contextual conversations that simulate expert guidance. The system architecture was developed through modular backend/frontend integration, and iteratively refined through usability feedback and internal testing cycles. A user study involving thirty participants at Arizona State University was conducted to evaluate the chatbot’s effectiveness. Results from post-interaction surveys showed high user satisfaction in areas such as response clarity, beverage recommendation accuracy, and conversational tone. Most users found the chatbot easy to use, educational, and engaging, while personalization features were well-received—though opportunities for refinement in response speed and interface clarity were identified. Updates made based on this feedback included onboarding instructions, improved preference visibility, and backend optimizations to reduce latency. This work demonstrates how generative AI models can be applied meaningfully in experiential learning contexts, particularly those requiring nuanced guidance and dynamic user engagement. The findings contribute to ongoing discussions about the role of large language models in education and present a scalable model for future AI-driven learning applications within lifestyle and hospitality domains. Chatbot Design User Experience Personalized Recommendations Beverage Education Response Time Onboarding human-computer interaction Survey Analysis Mobile App Evaluation React Native Render FastAPI Firebase Chatbot AI Interface Accessibility Sommelier Beverage Usability The Virtual Sommelier: Developing an AI Chatbot for Interactive Beverage Learning