Mind Map Automation
As part of the graduation requirements for Middle Tennessee State University’s Honors College, I got the opportunity to work on an honors thesis with a faculty advisor. This thesis was titled “Mind Map Automation: Using Natural Language Processing to Graphically Represent a Portion of a U.S. History Textbook.” The goal I had for this project was to show how NLP could be used to create a supplementary study tool for students.
Mind mapping is a technique for note taking where an individual jots down details that radiate from a centralized idea (see above for an example). In my thesis, I took sections of an open education resource, OpenStax’s U.S.History textbook, and preprocessed them using pdfplumber. The next task was to identify key concepts in the sections along with their corresponding properties using Python. This was done through a combination of NLP methods like coreference resolution, text summarization through BERT, and dependency parsing. The results of this approach were graphically depicted in a static webpage using the radial tidy tree example from the D3 JavaScript library.
Through working on this project, I gained more hands-on experience with Natural Language, data visualization, and open education resources. This project was also done prior to the many recent advancements in NLP and if I were to revist it, I would utilize newer machine learning tools and decelop a dynamic web application to display the results.
A full copy of my thesis, which dives further into the methodology and challenges of this project, can be found here on JEWLScholar. Additionally, all the programs and supplementary materials used for this project can be found in this GitHub repository.
