ISSR Methodology Workshop | Topic Modeling in R: LDA and Structural Topic Models

Instructor: Venkat Dasari

Topic modeling is a computational approach for identifying patterns across large collections of documents, including political speeches, news articles, interview transcripts, and open-ended survey responses. This workshop introduces participants to two widely used approaches in R: Latent Dirichlet Allocation (LDA) and Structural Topic Models (STM).

Through hands-on examples, participants will learn how to prepare text, estimate and interpret topic models, select the number of topics, and identify representative words and documents. We’ll also examine how STM incorporates document-level metadata to study variation in topic prevalence across groups or over time. Throughout the workshop, we’ll emphasize model evaluation, substantive interpretation, and reproducibility.

Learning Objectives:

By the end of this workshop, participants will be able to:

• Explain the basic concepts behind LDA and Structural Topic Models
• Prepare text data for topic modeling in R
• Estimate and interpret LDA and STM models
• Identify representative terms and documents for each topic
• Use document-level metadata to examine variation in topic prevalence
• Compare model specifications and select an appropriate number of topics
• Evaluate and present topic-modeling results using reproducible workflows

Prerequisites:

Prior experience with R is expected. Participants should be familiar with basic R commands, data frames, and working with libraries. Previous experience with text preprocessing, corpus objects, and document-feature matrices is strongly recommended. The ISSR workshop “Introduction to Text Analysis in R,” or equivalent experience with the quanteda package, provides suitable preparation. No prior experience with topic modeling, LDA, or STM is required.