Natural Language Processing of Short Comments from United States Navy Survey Data
Abstract
Invaluable information pertinent to decision making in naval planning and policy can be extracted from free response survey comments. However, processing survey comments for analysis can cost considerable time and funding depending on the methods used. We extend the work of Cairolis 2017 Naval Postgraduate School thesis, "Categorization of Survey Text Utilizing Natural Language Processing, and demographic filtering to aid the Navy in analysis of short-answer, free response comments from Navy surveys. Furthermore, we adopt similar approaches of text analysis from Layugs 2018 Naval Postgraduate School thesis, Extracting Major Topics from Survey Text Responses Using Natural Language Processing, in our efforts to discover meaningful topics. Through our newly modified text-mining methods, we aim to enhance the assignment process of survey comments to a particular topic. We apply our approach through analyzing comments from the Navy Exit Survey question, Why are sailors leaving? and the Navy Milestone Survey question, What will make sailors stay on active duty?. Our exploration of text mining processes includes implementing lemmatization, discovering topics using the relevancy metric Latent Dirichlet Allocation (LDA), creating a new comment-to-topic assignment process, and developing a web-based application that illustrates these methods.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2020
- Accession Number
- AD1114365
Entities
People
- Marvin P. Salonga
Organizations
- Naval Postgraduate School