AI-Powered Decision Support Systems in Supply Chain Management

Overview

This repository contains code and resources related to the research project on the integration of Large Language Models (LLMs) and AI-powered agent networks in Supply Chain Management (SCM). The project aims to explore the potential of LLMs in enhancing decision support systems, addressing current challenges, and identifying future research opportunities.

Research Abstract

The integration of Artificial Intelligence (AI), particularly Large Language Models (LLMs) and LLM-based agent networks, has the potential to revolutionize Supply Chain Management (SCM) by enhancing decision support systems and operational efficiencies. This research presents a comprehensive literature review on the application of LLMs in SCM, synthesizing current research and identifying critical challenges and opportunities. Key challenges identified include data integration, bias in LLM outputs, explainability, ethical considerations, security concerns, and scalability. Future research directions emphasize the need for exploring LLM-powered agent networks, particularly in urban supply chains and fresh food logistics.

Project Structure

The project is organized into several key components:

• data/: Contains datasets used for literature review and analysis.
• src/: Includes Python scripts for data retrieval, processing, analysis, and visualization.
• notebooks/: Jupyter notebooks demonstrating exploratory data analysis and key findings.
• app/: Contains the Streamlit application for interactive querying of the literature.
• requirements.txt: Lists the required Python packages for the project.

Key Features

• Literature Review: Comprehensive analysis of existing studies on LLM applications in SCM, identifying research gaps and opportunities.
• Keyword Analysis: Python scripts to extract and analyze keywords from publications using DOI lookup APIs.
• Clustering and Visualization: Clustering of publications based on abstracts using embedding models, with visual representations of clusters.
• RAG-Based Streamlit App: An interactive application allowing users to query a database of publications and receive context-aware responses powered by LLMs.
• Bibliometric Analysis: Statistical analysis of publication trends, citation counts, and co-authorship networks.