As the importance of diet in overall health management grows, access to accurate and personalized nutritional information becomes essential. Despite the abundance of recipes online, calculating detailed nutritional values remains challenging. Our research describes the construction of a system for recipe analysis, indexing, and question-answering that leverages a comprehensive ingredient Knowledge Graph (KG) to provide nutritional evaluations. The system has been developed in Greek and can be extended to support other languages. The system uses web crawling to collect recipes from the web in Greek, extracts the ingredients, and represents them using a KG in a Neo4j graph database. This KG is enriched with nutritional data for each ingredient from reliable sources. Recipes, ingredients, and their categories are represented as connected nodes in the KG, facilitating easy and effective Cypher query construction. Additionally, a machine learning model extracts user intent and entities about recipes and their nutritional values from Greek language queries. The extracted values are utilized for rule-based Cypher query creation, enabling users to query the database and obtain answers. The system is easily maintainable, allowing new ingredients and recipes to be added seamlessly. The proposed question answering system supports efficient information retrieval based on nutritional values calculations. It is particularly useful for dietary health management, where precise nutritional regimens are essential. The natural language interface makes the system accessible to a wide audience, including people without technical expertise, enabling easy inquiries about recipes and nutritional data. The system can also be incorporated into meal and diet planning, broadening its application in both personal and clinical dietary management.