Comparison

Ingredient substitution in recipes aims to change one ingredient with another one. This may help people exploring culinary practice, personalize their recipe so that it can correspond to their dietary needs, and help them avoid potential allergens. To help people substitute ingredients in recipes, many AI technologies are proposed. The following table compares several ingredient substitutions models. These models are composed of statistical-based models and language models-based models. This table shows that the model composed of the context encoder, ingredient decoder and ingredient substitution decoder modules outperforms the baselines on all metrics.