Named Entity Recognition using LLMs – April 16, 2025
Session Description
April 16 2025 @ 11:30 am - 1:00 pm
Over the past few years, large language models (LLMs), like ChatGPT and Claude, have become instrumental in performing various natural language processing (NLP) tasks. They are proficient in natural language understanding and can produce quality responses to various NLP problems. Despite this, LLMs need to be guided and prompted in specific ways to have optimized and consistent outputs.
In this course, we will learn how to leverage LLMs to classify named entities in texts through zero-shot, single-shot, and few-shot classification. We will learn how to ensure consistent structure in our outputs via OpenAI’s new line of GPT-4o models and Pydantic. Most importantly, we will learn how to leverage these outputs to frame and answer humanities-specific qualitative and quantitative questions.
This course is taught by William Mattingly and will run on:
- Monday, April 14, 2025: 11:30am – 1:00pm ET
- Wednesday, April 16, 2025: 11:30am – 1:00pm ET
- Friday, April 18, 2025: 11:30am – 1:00pm ET
Instructor: William Mattingly
William Mattingly is a postdoc in the Smithsonian Institution’s Data Science Lab, where he develops machine learning and natural language processing pipelines for processing archival records. He is a Harry Frank Guggenheim fellow for his work as co-PI on the Bitter Aloe Project. He is also co-PI for the ACLS-funded Personal Writes the Political project and NLP engineer and data scientist for the NEH – funded Placing the Holocaust project. His publications include a textbook that teaches Python to humanists and the ethical application of machine learning on archival data.
