An Active Learning Approach to Transcript Coding for Education Research

Educational research relies on a labor intensive analysis of qualitative data. Many times, data takes the form of transcribed conversations between students and teachers. These transcripts are analyzed, annotated, and coded to help uncover patterns and ideas. An education researcher will comb through the documents, line by line, classifying each line or paragraph into a pre-described category. This paper presents the framework of a new tool that has the potential to greatly aid education researchers as they analyze qualitative data. By utilizing statistical machine intelligence techniques for Natural Language Processing (NLP), and leveraging existing Deep Learning language models, the tool completes much of the grunt work of transcript coding, letting the researcher focus his or her attention on the areas that are important.