The need to provide reliable information in healthy food recommendations to consumers has become more imperative as misinformation continues to spread and obesity rates continue to rise. When consumers search for recipes, they are often interested in the nutritional information. The problem is much of this nutritional information can be difficult to interpret if the user is not a nutritionist or some form of food expert. For example, if a recipe lists 12g of sugar per serving, how will a normal consumer with no nutritional background have insight into what this means for their health. Is 12g of sugar per serving healthy or not and if it is healthy, are there underlying conditions such as Diabetes that would suggest the recipe is unhealthy for that specific condition? Having a platform that can distill this information and provide recipe healthiness to a user and also recommend to them healthy meals based on their food and individual preferences could have major use cases. For starters, consumers would be able to better interpret which recipes are healthy and which are not. The consumer would receive healthier recommendations for their meals which could lead to an improved lifestyle and also lower risks of dietary diseases. While there are methods to estimate the healthiness of online recipes in the presence of nutritional informational, there may be occasions when nutritional information is not available. Also if the nutritional information is provided, the user may still not comprehend the meaning behind the values. Utilizing relationships between recipe ingredient information would enable us to solve this problem and allow us to accurately recommend healthy and preferable meals to consumers. By considering only healthy recipes to be recommended to users, a recommendation system can take a graph of users and their preferences/demographic information and train a factorization model based on this graph of users to be utilized for providing incoming users with healthy and desirable meal recommendations.