The Classification of Documents Using the Subspace Relevance Model

Open Access
Author:
Henry, Patrick A
Area of Honors:
Computer Science
Degree:
Bachelor of Science
Document Type:
Thesis
Thesis Supervisors:
  • Daniel Kifer, Thesis Supervisor
  • Jesse Louis Barlow, Honors Advisor
  • Jesse Louis Barlow, Faculty Reader
Keywords:
  • Document
  • Clustering
  • Classifcation
Abstract:
The document classification problem is one that identifies how we can efficiently take documents about various subjects and automatically assign them to a specific category. This paper considers the subspace relevance model as a way of classifying documents. Using the steepest descent method over the Stiefel and Grassmannian manifolds, we are able to find the local minimum of a specified function. Our goal is to create an algorithm that will perform document classification efficiently and effectively.