FoodBot@EdUHK is a chatbot prototype designed around the practical question of what to eat on campus. It brings together EdUHK dining information, including menus, opening hours and nutritional or allergen-related records, so users can ask about options that fit preferences such as vegetarian, halal, high-protein or particular dietary restrictions.
The project uses a tailored knowledge base rather than a general restaurant search. Users can describe what they are looking for and receive meal suggestions or dining information in a conversational format. The report also documents the care needed when working with food data: prompt settings, model choice and data cleaning were tested to reduce unsuitable recommendations. This is a student project prototype and should not replace official allergy, ingredient or dietary advice from food providers.
What FoodBot brings together
- Conversational search for campus meal options
- Dining information including menus, opening hours and nutritional records
- Preference prompts for vegetarian, halal, high-protein and dietary-restriction queries
- A campus-specific food and allergen knowledge base
- Data-cleaning and prompt refinements tested against meal-recommendation cases
Findings and cautions
The report describes a working PoE-based chatbot and a dataset of around 1,900 food-related entries. It also records testing across multiple base models and settings; information accuracy remains a necessary limitation for any future public use.