Topic Modeling via Latent Dirichlet Allocation

Topic modelling is increasingly used by psychologists in all areas to make quantitative inferences from large bodies of text, whether it’s open-ended responses, interview transcripts, social media posts, or journal articles. In this workshop, you will receive an introductory overview of latent Dirichlet allocation (LDA), a popular method of topic modelling that identifies latent trends in subject matter based on visible patterns in term frequency in a corpus of many text documents. I will first provide a conceptual overview of topic modelling and LDA, and weigh the pros and cons of different topic modelling methods. Then, I will discuss how to create a topic model using LDA, addressing software possibilities and the necessary components of data selection, pre-processing, and analysis. This workshop will be of interest to researchers with no experience using LDA, hoping to make sense of the glut of textual data at their disposal.