Every year the amount of recipes published on the internet is increasing fast. People around the world want to share their culinary experiences with other people. This trend creates various recipes for a user and makes cooking more interesting and fun. However, the process of choosing a recipe among all recipes becomes a difficult task for the reader. By reading the recipe it is hard to define the difficulty level of it. It can cause a problem during the cooking process. As a result, users can be disappointed in the cooking process. To prevent such problems it is better to know the difficulty level of the recipe. But most recipe-related services do not include the categorization by difficulty level.

In this paper, we propose using the graph neural network to create a recipe evaluation system that classifies meal recipes according to the difficulty level.