Research
My research spans several exciting areas within computer science, including AI for Software Engineering, Human Factors in Software Engineering, Cybersecurity, and Generative AI. Below are some of the projects I am currently working on.
Automatic Software Documentation
Understanding code can be a daunting task, often requiring extensive time and effort. To address this, I've developed a tool, called COMCAT, that leverages advanced AI to automatically generate comments for code. By intelligently identifying key sections of code and generating relevant comments, this tool helps developers understand and maintain software more efficiently. This project combines cutting-edge AI with practical software engineering, making code comprehension easier and faster.
Automated Assessment of Programming Assignments
Grading programming assignments in large computer science classes can be challenging and time-consuming. My research focuses on using AI, specifically ChatGPT-4, to automate and enhance the assessment process. By developing methods to accurately evaluate code for correctness, style, and efficiency, this project aims to provide consistent, scalable, and unbiased feedback to students, ultimately improving the educational experience.
Cognitive Biases in Cybersecurity
Cybersecurity is critical in protecting our digital lives. My research explores how cognitive biases can be used to influence and disrupt potential attackers. By embedding subtle psychological cues into web applications, we aim to mislead and hinder attackers' progress. This innovative approach seeks to enhance cybersecurity measures by turning attackers' own cognitive tendencies against them, making our digital environments safer and more secure.
Cybersecurity Education
Cybersecurity education is essential for preparing the next generation of professionals to protect our digital world. My research involves conducting human subjects studies with eye tracking technology to gain insights into what topics novice hackers find difficult. Additionally, I explore the differences in learning strategies between successful and unsuccessful learners. This work aims to improve educational strategies and materials, ultimately enhancing the effectiveness of cybersecurity training programs.
Reverse Engineering Tool Metrics
Reverse engineering is a necessary tool in defenses against emerging cybersecurity threats like sophisticated and stealthy malware. Many modern approaches use AI tools to simplify decompiled code and recover information like function and variable names and types. These tools are measured by metrics like BLEU scores to determine how close their recovery resembles the original information from the source code. My research into this topic involves determining how useful these metrics are, as well as investigating whether higher scores in these metrics correlated with improved performance on real reverse engineering tasks involving decompiled code. If these metrics are not useful, then we will need to develop better methodologies for developing and testing these tools.