Ai based recipe generation by integrating ingredient pairing score determined from flavor spaces

A system, comprising:

• a memory storing instructions;
• one or more communication interfaces;
• and
• one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to:
• determine a set of ingredients from a plurality of ingredients based on a set of macronutrient profiles suitable for one or more medical conditions of a user;
• determine a set of secondary ingredients from the set of ingredients by evaluating an ingredient compatibility between each ingredient amongst the set of ingredients with a primary ingredient selected from the set of ingredients, wherein the ingredient compatibility is evaluated based on an ingredient pairing score calculated for each ingredient amongst the set of ingredients using a directional alignment measure, an orientation factor, and a spatial proximity measure determined from a molecular embedding matrix represented as a flavor space;
• and
• generate a recipe comprising a set of cooking instructions from the primary ingredient and the set of secondary ingredients using a text generation model fine-tuned on a recipe dataset based on an optimal token length estimated from an optimal token count per cooking instruction.

One or more non-transitory machine-readable information storage mediums comprising one or more instructions which when executed by one or more hardware processors cause:

• determining a set of ingredients from a plurality of ingredients based on a set of macronutrient profiles suitable for one or more medical conditions of a user;
• determining a set of secondary ingredients from the set of ingredients by evaluating an ingredient compatibility between each ingredient amongst the set of ingredients with a primary ingredient selected from the set of ingredients, wherein the ingredient compatibility is evaluated based on an ingredient pairing score calculated for each ingredient amongst the set of ingredients using a directional alignment measure, an orientation factor, and a spatial proximity measure determined from a molecular embedding matrix represented as a flavor space;
• and
• generating a recipe comprising a set of cooking instructions from the primary ingredient and the set of secondary ingredients using a text generation model fine-tuned on a recipe dataset based on an optimal token length estimated from an optimal token count per cooking instruction.