Dietary Assistant Powered by Language Models and Knowledge Graphs

The primary goal of this project is to develop a dietary assistant utilizing a large language model (LLM) interfaced with a comprehensive knowledge graph of food-to-chemical relationships. This assistant aims to provide users with dietary recommendations, potential allergen identification, and nutritional information, enhancing their dietary choices based on scientific data. The specific aims are (1) development of a dietary assistant chatbot that leverages large language models (LLMs) integrated with a comprehensive knowledge graph, FoodAtlas, and that offers detailed nutritional information to users, thereby enhancing their dietary choices based on scientific data, (2) add an explainability module to the LLM for understanding decisions, and (3) expand the dietary assistant to become an “agent” that can act as a dietician for creating a personalized and holistic dietary regiment.