Event Type Classification and Trend Analysis of Cybersecurity News Using Dynamic Topic Modeling

In the present interconnected era, where social media, online infrastructure, cloud computing systems, and automated processes have become integral to our daily lives, technology is advancing rapidly. However, this progress brings a corresponding rise in cybersecurity incidents. Attackers are continuously adapting, employing new attack methods, tools, and strategies that enable them to breach even the most sophisticated and tightly controlled environments. This poses a greater risk of significant damage or covert operations going unnoticed. The proliferation of data and interconnected devices further expands the potential targets for such attacks. Therefore, it is crucial to allocate more resources and attention to cybersecurity, especially in vulnerable sectors.The cost of cybersecurity attacks is expected to increase by a factor of 3.5 times over a ten-year period from 2015 to 2025. This substantial increase far surpasses the economic repercussions brought about by most natural disasters within a single year. This occurrence not only signifies the largest shift of economic resources in history but also presents a substantial threat to the driving forces behind innovation and investment.In this praxis, we conducted research focused on developing predictive models to classify cybersecurity text corpus into different event types. Our objective was to showcase the effectiveness of cutting-edge deep learning techniques and Large Language Models (LLMs) such as Bidirectional Long Short-Term Memory (Bi-LSTM) and Bidirectional Encoder Representations from Transformers (BERT) in assisting organizations in proactively addressing evolving cybersecurity incidents. Additionally, we employed Dynamic Embedded Topic Models to offer cybersecurity teams insights into emerging trends by identifying latent topics and word probabilities over time. The results from this praxis shows that the proposed models plays a vital role in assisting various important stakeholders within the cybersecurity domain, such as cybersecurity engineers and information officers. They provide valuable support by accurately classifying vast amounts of data and uncovering significant trends. This, in turn, empowers these stakeholders to make proactive decisions and implement robust cybersecurity strategies.

Author

• Akinfaderin, Adewale Abiodun

Language

• en

Date created

• 2023

Type of Work

• Dissertation

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Engineering Management

Degree

• D.Engr.

Advisor

• Etemadi, Amir

Committee Member(s)

• Yassan, Rebecca
• Widener, Donald

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/hd76s087q