Nutrition Sources & Methodology

1. How Koben Estimates Nutrition

Koben provides estimated and approximate nutrition values based on recipe information you create or import, including ingredient names, amounts, units, and servings.

• Recipe text is parsed into structured ingredient entries.
• Ingredient names and units are normalized to common food and serving formats.
• AI-assisted extraction helps identify ingredients and quantities from incomplete or unstructured recipe text.
• An OpenAI large language model is used to generate approximate calories and macro estimates from the parsed recipe content and serving assumptions.

If imported or user-provided recipe content is missing, ambiguous, or imprecise, the resulting values may be less accurate.

2. Sources & References

Koben's current nutrition estimates are generated with OpenAI models from user-provided or imported recipe content. The references below describe the model platform and safety context for these estimates.

• Unveiling the accuracy of ChatGPT's nutritional estimations
• An Evaluation of ChatGPT for Nutrient Content Estimation from Meal Photographs
• OpenAI Usage Policies
• OpenAI Terms of Use

3. Limitations

• Estimated values can differ from branded packaging labels and lab-tested nutrition data.
• Imported social media recipes may be incomplete, informal, or imprecise.
• Portion-size assumptions can materially change calories and macro totals.

4. Medical Safety Disclaimer

Nutrition information in Koben is provided for informational and educational purposes only. It is not medical, nutrition, or treatment advice. Consult a qualified healthcare professional before making medical decisions.

5. Contact & Corrections

If you notice issues with nutrition estimates or extracted ingredient data, contact imeri.soft@gmail.com.