CSH Talk by Jaber Fooladi (ASCII)

In this talk, Jaber Fooladi will present their methodology for efficiently processing large volumes of news data using fine-tuned local language models (LLMs).

Their approach focuses on extracting critical information relevant to supply chain networks, specifically identifying supply-buyer and partnership relationships between companies and countries.

They demonstrate how they construct knowledge graphs from this extracted information to visualize and analyze complex business relationships clearly and effectively.

Additionally, they discuss techniques for monitoring company behaviors, such as trade activities, over time.

Finally, they address methods for comparing real-world trade flows, with a particular emphasis on industry-specific analyses, illustrated through a case study of the steel industry’s trade flows as derived from news-based trade information.