Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Enhancing FKG.in: automating Indian food composition analysis

Created by
  • Haebom

Author

Saransh Kumar Gupta, Lipika Dey, Partha Pratim Das, Geeta Trilok-Kumar, Ramesh Jain

Outline

This paper presents a novel method for computing food composition data for Indian recipes using the Indian Food Knowledge Graph (FKG[. ]in) and a large-scale language model (LLM). We provide an overview of an automated food composition analysis workflow, focusing on its core functions: nutritional data aggregation, food composition analysis, and LLM-based information resolution. This workflow complements FKG[. ]in and aims to iteratively enrich food composition data from a validated knowledge base. We also highlight the challenges of representing Indian food and accessing food composition data digitally, examining three major food composition data sources: the Indian Food Composition Chart, the Indian Nutrition Databank, and the Nutritionix API. We briefly describe how users interact with the workflow to obtain diet-based health recommendations and detailed food composition information for various recipes. We explore the complex challenges of analyzing Indian recipe information across multiple dimensions, such as structure, multilingualism, and uncertainty, and present ongoing research on LLM-based solutions to address these challenges. The proposed AI-based method for knowledge curation and information resolution is application-agnostic and can be replicated across domains.

Takeaways, Limitations

•
Takeaways:
◦
A novel workflow for automated analysis of food composition data from Indian recipes is presented.
◦
Proposing an efficient data aggregation and analysis method using FKG[. ]in and LLM.
◦
Potential to improve data accuracy through the use of diverse data sources and iterative supplementation of knowledge base.
◦
Presenting a generalizable methodology that is application-independent.
◦
The possibility of providing diet-based health recommendations.
•
Limitations:
◦
A complete solution to the complexities of Indian cuisine, including its structure, multilingualism, and uncertainty, is still incomplete.
◦
Further validation of the performance and reliability of LLM-based solutions is needed.
◦
The difficulty of obtaining comprehensive data on the diverse Indian cuisine and regional differences.
◦
Reliability and consistency issues with data sources.
👍
Made with Slashpage