Date of Award
Open Access Thesis
Master of Science (MS)
We propose a neural network model for classifying bubbles (circles) used in instructor course evaluations. The model is trained on prior (labeled) objects consisting of bubbles and general text. The trained model is then used to determine the positions of bubble answer options on a given evaluation form. A Web portal accompanies the classification system and facilitates management of the network and analysis of the results. The departmental staff will upload an unevaluated form per course and the system will execute the neural network model on it; application logic will be responsible for ensuring data persistence of the bubble positions in addition to student long-form question answers. Once the departmental staff uploads an electronic copy of the filled out evaluations for a course into the portal, the application server will aggregate the results based on the output from the neural network. The instructor for the course is able to view the evaluation results once granted access by the departmental staff.
Held, Jason, "A Neural Network Model for Classifying Bubble-Based Instructor Evaluations, and an Accompanying Web Portal" (2018). Graduate Masters Theses. 492.