Large Language Models for Social Science Research

About

Workshop Contents and Objectives

The course aims to bring students and researchers in the social sciences up to speed with the technology of Large Language Models (LLMs). LLMs are a powerful machine learning technique able to represent and generate text, as well as data in other modalities. LLMs can also be used for automating several research tasks with little tuning. The course provides a theoretical foundation on LLMs as well as several applications relevant to the social sciences, which will be implemented in practical sessions.

The practical sessions will showcase examples from recent literature in the social sciences using LLMs in a variety of novel ways. These include the large-scale detection of online media polarization, the thematic clustering and labeling of textual sources, how to simulate survey participants, and how to use LLMs as personal research assistants for searching and summarizing information, or for data analysis.

Workshop design

The course combines lectures (morning sessions) with laboratories consisting of live coding and practicals (afternoon sessions). The students will also be provided with a small amount of credits to use a commercial LLM during the course.

Detailed lecture plan (daily schedule)

Day 1
Introduction to Large Language Models

Laboratory: System setup, the HuggingFace library, using the OpenAI API.

Day 2
LLMs for content representation

Laboratory: Thematic clustering of textual sources.

Day 3
LLMs for classification and regression

Laboratory: Large-scale, zero-shot classification of textual sources.

Day 4
LLMs for content generation

Laboratory: Prompt engineering and simulating survey participants.

Day 5
LLMs as research assistants

Laboratory: Semantic search, retrieval-augmented generation, data analysis.

Course feedback and Q&A.

Class materials

All materials will be provided online in the course’s GitHub repository.

**The Summer School cannot grant credits. We only deliver a Certificate of Participation, i.e. we certify your attendance.**

If you consider using Summer School workshops to obtain credits (ECTS), you will have to investigate at your home institution (contact the person/institute responsible for your degree) to find out whether they recognise the Summer School, how many credits can be earned from a workshop/course with roughly 35 hours of teaching, no graded work, and no exams.

Make sure to investigate this matter before registering if this is important to you

Course leader

Giovanni Colavizza is a Professor of Computer Science at the Department of Communication, University of Copenhagen, and Associate Professor of Computer Science at the Department of Classical and Italian Philology and the Digital Humanities Advanced Research Centre of the University of Bologna

Target group

graduate students, doctoral researchers, early career researchers

Prerequisites: practical knowledge of the Python programming language is recommended

Fee info