Date of Award

12-2024

Degree Type

Thesis

Degree Name

M.S.

Degree Program

Computer Science

Department

Computer Science

Major Professor

Dr. Md Tamjidul Hoque

Second Advisor

Dr. Abdul Rahman Alsamman

Third Advisor

Dr. Ben Samuel

Fourth Advisor

Dr. Shreya Banerjee

Fifth Advisor

Dr. Christopher Summa

Sixth Advisor

Dr. Atriya Sen

Seventh Advisor

Dr. Phani Vadrevu

Abstract

In the digital age, text-based passwords remain a primary method for securing online accounts. Yet, users frequently face a dilemma between creating passwords that are easy to remember and sufficiently secure against cyberattacks. This research introduces an approach to password generation that bridges this gap by utilizing linguistic patterns, particularly song lyrics, to develop highly secure and naturally memorable passwords. Using large lyric datasets gained from web scrapes from popular song lyric websites (AZ Lyrics, Genius), features are extracted from a corpus of over 5 million lyrics using sentence structure and natural language processing in a novel way. In using transformer architecture, we automate the process of generating password phrases based on these features, ensuring a balance between complexity and ease of recall. These generations are evaluated in a user study, given the subjective nature of language memorability. Our system evaluates password strength using LSTM-layered recurrent neural network models that assess the likelihood of successful password cracking attempts, and we provide users with memorability aids, such as narrative cues or mnemonic devices, using large language models to enhance long-term retention. The machine learning models for password security are evaluated using validation and test splits, as well as cross-validation, and compared analytically. These tools are integrated into a user-friendly interface designed to educate users on best practices while simplifying the process of creating and managing passwords. This approach to using lyric-based features for generating passwords is both generalizable to other pieces of literature and novel in its application using machine learning. Similar work has generated sequences from lyrics and analyzed security of passwords, however generation has not been done using machine learning and the combination of these applications has not been realized.

Drawing from cognitive science, the research demonstrates that familiar linguistic structures can enhance password recall without compromising security. We compare traditional password generation methods to our machine learning-based approach through user studies, revealing improved usability and security. This study offers a forward-thinking method that redefines how passwords can be both secure and user-friendly, enhancing overall cybersecurity while addressing the common issues of password fatigue and memory overload.

Rights

The University of New Orleans and its agents retain the non-exclusive license to archive and make accessible this dissertation or thesis in whole or in part in all forms of media, now or hereafter known. The author retains all other ownership rights to the copyright of the thesis or dissertation.

Download

Share

COinS