The UTC Graduate School is pleased to announce that Ayman Gumaa will present Master's research titled, Applying KGQA to Cybersecurity on 06/11/2024 at 10:30 am - 12:00 noon at https://tennessee.zoom.us/my/mengjunxie. Everyone is invited to attend. 

Computer Science

Chair: Dr. Mengjun Xie

Co-Chair: 

Abstract:

Knowledge Graph Question Answering (KGQA) systems. especially UTC's ForensiQ, show remarkable results when applied to cybersecurity data. However, a lot of effort is needed to explore a custom KG, use a KGQA system to answer questions using it, generate KGQA datasets for it, and understand what factors affect this system performance and why it produced this answer for this question. We created a web application so all of these tasks can be done more easily using this single application. We also implemented a new entity detection approach for ForensiQ that reduced inference time by an order of magnitude while maintaining ForensiQ's accuracy. Finally, we rephrased the question templates so test questions differ from training questions in their choice of words and sentence structure. Our results confirmed that both ForensiQ models perform the same and outperform other baseline systems.