Natural Language Processing and Student Reflective Writing
Project Author
Issue Date
2024-05-03
Authors
Abraham, Michael
Loading...
Embargo
First Reader
Luman, Douglas J.
Additional Readers
Hart, D. Alexis
Keywords
Natural Language Processing , NLP , Data , Data Collection
Distribution
Abstract
The goal of this thesis is to improve technical writing through the development of a data-trained model that is able to score student reflective writing (SRW) based on an established standard. Student reflection most closely reflects technical writing in the Computer Information Science department, so the model focuses on the improvement of SRW. Technical writing is an aspect of Computer Science that is fairly absent from the department, so instead of creating an entirely new target of research, it would be useful to utilize the closest form of technical writing in the department. Defining what good technical writing consists of meant that I had to use literature to come up with a definition/standard that I would then implement with my model. Defining good technical writing also meant acknowledging the positives and negatives that come with technical writing. Once exploring technical writing and its intricacies, I developed the model that would use my research as a basis for scoring responses based on a crafted rubric. The model was built by utilizing Natural Language Processing (NLP), specifically the Python library Gensim and its method for word embedding known as word2vec. By exploring the capabilities of different topic models and word embeddings I assert that word2vec modeling would be the best method for utilizing information retrieval in this context. Through this research, a system for improving technical writing was revealed; this system utilizes word2vec processing, proselint, and syntax tree. However, implementation of this system was unable to be entirely carried out due to time issues that arose in the development of the thesis. Future steps in developing the project further and implementing systems that would allow the project to be fully automated are described in the thesis.
Description
Collections
Chair
Major
Computer Science
English Major-Emphasis in Literature
English Major-Emphasis in Literature
Department
Computer and Information Science
English
English
Recorder
License
Citation
Version
Honors
English, 2024