• Home
• Issues
• 2097

LLM Framework for Enhanced Supply Chain Data Analysis and Decision Support: A Comparative StudyLLM Framework for Enhanced Supply Chain Data Analysis and Decision Support: A Comparative Study

Select Volume / Issue:

• Volume 13, Issue 3, May 2026

Year:

2026

Type of Publication:

Article

Keywords:

LLM, Data Analysis, Decision Support, Comparative Study

Authors:

• Cho, Johan
• Zeid, Abe

Journal:

IJIRES

Volume:

13

Number:

3

Pages:

39-59

Month:

May

ISSN:

2349-5219

BibTex:

@article{2097, author = "Johan Cho and Abe Zeid", abstract = "Current supply chain management techniques rely on rigid business intelligence systems such as PowerBI and Tableu that do not provide intuitive natural language insights from complex, multi-dimensional datasets. These systems typically require a heavy time and monetary investment plus specialization and domain-specific knowledge to build and use. Supply chains generate vast quantities of mutli-level and often unorganized data, spanning inventory levels, supplier performance, transportation metrics, and quality control results. Thus, a gap between data availability and actionable insights and information forms. This paper presents a Large Language Models (LLMs) framework that uses specialized AI agents (OpenAI, Gemini, DeepSeek) to provide supply chain data analysis through natural language interfaces. This framework incorporates model selection, chat history management, and data analysis capabilities though retrieval-augmented generation (RAG) that combine to create a specialized supply-chain-specific assistant. Through evaluation on real supply chain datasets, this LLM system demonstrates general analytical accuracy, acceptable response quality, and improved cost-effectiveness compared to generalized LLMs. It also solves many of the problems of traditional supply chain analysis while also creating a promising opportunity to enhance the field. This system validates this approach as a promising application of LLMs within supply chain management and other analysis-focused fields.", issn = "23495219", journal = "IJIRES", keywords = "LLM, Data Analysis, Decision Support, Comparative Study", month = "May", number = "3", pages = "39-59", title = "{LLM} {F}ramework for {E}nhanced {S}upply {C}hain {D}ata {A}nalysis and {D}ecision {S}upport: {A} {C}omparative {S}tudy{LLM} {F}ramework for {E}nhanced {S}upply {C}hain {D}ata {A}nalysis and {D}ecision {S}upport: {A} {C}omparative {S}tudy", volume = "13", year = "2026", }

Abstract:

Current supply chain management techniques rely on rigid business intelligence systems such as PowerBI and Tableu that do not provide intuitive natural language insights from complex, multi-dimensional datasets. These systems typically require a heavy time and monetary investment plus specialization and domain-specific knowledge to build and use. Supply chains generate vast quantities of mutli-level and often unorganized data, spanning inventory levels, supplier performance, transportation metrics, and quality control results. Thus, a gap between data availability and actionable insights and information forms. This paper presents a Large Language Models (LLMs) framework that uses specialized AI agents (OpenAI, Gemini, DeepSeek) to provide supply chain data analysis through natural language interfaces. This framework incorporates model selection, chat history management, and data analysis capabilities though retrieval-augmented generation (RAG) that combine to create a specialized supply-chain-specific assistant. Through evaluation on real supply chain datasets, this LLM system demonstrates general analytical accuracy, acceptable response quality, and improved cost-effectiveness compared to generalized LLMs. It also solves many of the problems of traditional supply chain analysis while also creating a promising opportunity to enhance the field. This system validates this approach as a promising application of LLMs within supply chain management and other analysis-focused fields.

Full text: IJIRES_2097_FINAL.pdf Online version [Bibtex]