NEXT-GENERATION FOOD RECOMMENDATION SYSTEM: A REAL-TIME, FEEDBACK-DRIVEN CHATBOT SOLUTION FOR RESTAURANTS

Abstract

The growing demand for personalized dining experiences has driven the develop- ment of intelligent food recommendation systems in the restaurant industry. This pa- per presents a novel chatbot solution that integrates Groq’s ultra-fast inference engine, Meta’s LLaMA 3–8B model with an 8192-token context window, and Firebase’s real- time database to deliver dynamic, diet-aware menu suggestions. Customer feedback, menu items, and ingredient lists are stored and continuously updated in Firebase, while users communicate their dietary needs directly to the chatbot at runtime. The chatbot interface interprets user queries and dietary preferences on the fly, applies sentiment analysis to refine dish popularity metrics, and evaluates ingredient compatibility to generate tailored recommendations (e.g., gluten-free, vegan, low-calorie). Ingredient- level transparency enables users to make informed choices, while the system’s real-time learning loop adjusts suggestions instantly based on new feedback. Preliminary evalu- ation indicates significant improvements in user satisfaction and operational efficiency compared to static or batch-updated recommender models.  The proposed solution bridges existing gaps by offering personalized, interactive, and up-to-date food recom- mendations, demonstrating its potential to enhance both customer engagement and restaurant service quality.