Speaker: Professor Sebastian Muller (Technical University of Munich School of Management)
Hosted by: University of Liverpool Management School's Accounting and Finance Group
Open to: all University of Liverpool staff and students, with no sign up needed
Date: Wednesday 11 October 2023
Time: 15:00 - 16:15
Place: Management School - Seminar Room 1
Abstract
Applying large language models (GPT-3, Luminous, T5-XXL) to business descriptions of more than 79,000 firms, we construct business networks (BNs) of economically linked firms across the globe.
We run multiple evaluation tasks and find that our networks are better suited to identify relevant competitors, suppliers, and customers across the globe than traditional industry classifications.
We further demonstrate the usability of our networks by examining international lead-lag effects and M&A activities.
Our findings highlight the importance of considering both domestic and foreign economic links across the globe.
Co-authored by Professor Sebastian Muller and Christian Breitung.
Keywords
Business Network, Textual Analysis, Natural Language Processing, Transformer, GPT-3, Large Language Models
Full paper
Breitung, C. and Muller, S. (2023). 'Global Business Networks'.
- DOI: 10.2139/ssrn.4395079
Back to: Management School