Understanding Textual Data in Social Science Using Latent Dirichlet Allocation Model

• In Event: Innovative Approaches to Modeling Educational Data

Abstract

This study aims to show the applicability of LDA model to analyze the textual data of social science field. 198 abstracts from two journals, the Review of Educational Research and Psychological Bulletin, are analyzed. Based on top-20 words and top-10 representative documents, 4 topics were interpreted as “Genetic, or environmental factors causing individual differences”, “A way to improve the students’ academic achievements regarding the specific programs, or interventions”, “Explanations to cognitive-psychological concepts and relevant theories”, and “The issues about the person’s behaviors.” And we figured out that the interpretation of other abstracts with mixture of topics was the same the combination of the interpretations of each topic together.

Authors

• Minju Hong, University of Arkansas at Fayetteville

• Sunyoung Park, California Lutheran University