INFO 290

Cultural Analytics

Machine Learning and Measurements of Culture

Instructor: Kent Chang (kentkchang@berkeley.edu)

University of California, Berkeley, School of Information

Fall 2025: Tu/Th 11:00–12:30, South Hall 205

Syllabus Course catalog Schedule and topics

Course Description

While often defined as “the computational study of culture,” cultural analytics might be best understood as a radical interdisciplinary experiment, one that seeks to understand cultures—socialties, histories, cognition, the literary—through empirical models and patterns, built on effective computational representations of relevant cultural constructs. This experiment calls for a unique skillset: one needs to be familiar with approaches in the interpretive humanities and computer/information science; one also needs to cultivate an interdisciplinary mindset: recognize and appreciate the affordances and limitations of both qualitative and quantitative traditions. This class is imagined as a possible point of departure for those who are so inclined.

As such, this course is all about making connections: bridging the interpretive traditions of literary and cultural studies—which guide our critical engagement with texts and cultural artifacts—with computational methods, ranging from featurized classifiers to large language models. It pursues two complementary ends: students will develop interpretive strategies and critical vocabularies for cultural analysis, and, through hands-on practice grounded in engagement and experience with the text, they will learn to represent cultural data—whether text, image, audio, or video—and train machine learning models as algorithmic instruments to systematically characterize cultural phenomena of interest.

This class welcomes a range of inclinations: maybe you know how to implement an RNN or Transformer from scratch but are curious whether those models can be used to study literature, culture, or something beyond positive or negative sentiments. Or, maybe you’ve experimented with off-the-shelf topic models or word embeddings to explore humanities questions and want to see how far you can take them.

You can find the latest syllabus here. It is subject to change (can lean more humanistic or technical), depending on the background and interests of enrolled students.

Frequently Asked Questions

This class is designed for students interested in connecting interpretive approaches with computational tools, especially across the humanities and computer science. Broadly, this includes:
  • Students from the humanities who want to develop stronger intuitions for machine learning (ML) and natural language processing (NLP), and explore how they might apply these methods to their own work.
  • Students with some prior experience in ML and NLP who are interested in analyzing cultural and literary artifacts as real-world data—especially for questions relevant to the humanities.

As defined in this class, cultural analytics is an interdisciplinary practice that brings together critical theory and computational methods—particularly from natural language processing and computer vision—to study cultural phenomena such as literature and film. It explores how cultural artifacts can be represented as data, and how those representations support both interpretation and empirical analysis. Technically, it draws on approaches from natural language processing (including word embeddings, topic models, sequence modeling, and multimodal architectures), often aligning with the subfield of narrative understanding in NLP. At the same time, it remains grounded in humanistic inquiry, shaped by traditions in literary theory, philosophy, and cultural criticism. This class invites experimental work that measures, models, and interprets culture across disciplines. For more examples, see the Journal of Cultural Analytics and the 2024 Conference on Computational Humanities Research (CHR).

Graduate students can enroll directly through CalCentral. Undergraduate students must request a permission code (email Kent). The course is officially offered by the School of Information, but you do not need to be a student in the School to join.

This is a new class with a temporary course number (290s), so usually someone (me or you) has to petition or otherwise make a case for it to count. It has been approved as a technical elective for the Designated Emphasis in New Media. Reach out if you're interested, and we’ll see what we can do.

The final project constitutes the most significant portion of your grade. Given the range of approaches in cultural analytics, your project can lean more theoretical (or qualitative) or more computational—as long as it focuses on measuring some aspect of culture. You’re encouraged to draw on your strengths, whether in coding, close reading, or both.

See the prerequisites section of the syllabus—but the course is designed to be interdisciplinary and largely self-contained. If most students have strong close reading skills, we may condense relevant lab sessions. If most are experienced with deep learning, we’ll adjust discussion accordingly.

Feel free to reach out to Kent at kentkchang@berkeley.edu.

Updates

  • May 1, 2025: course website up