ChatBot for Cyber-Security based on RASA Framework

Abstract

With the ever-increasing complexity of cybersecurity threats and vulnerabilities, the need for efficient and effective vulnerability management solutions tends to be vital. This proposed cybersecurity chatbot seeks to introduce a more user-friendly way of getting insights about multiple kind of threats. This thesis explores the development and implementation of a novel chatbot leveraging Natural Language Processing (NLP) techniques for cybersecurity vulnerability searches. The implemented chatbot offers diverse functionalities, ranging from regular generic dialogues to in-depth exploration of cybersecurity threats. Users can inquire about threat levels, including low-threat vulnerabilities and zero-day threats. Users can explore MISP data, retrieving specific events and their correlations. The chatbot provides comprehensive details about specific threats like trojans, spyware, viruses, and adware, including functionality and many more insights. The chatbot facilitates comprehensive threat analysis by enabling users to filter threats based on date ranges, risk levels, and severity. The findings demonstrate the potential of chatbots as a valuable tool for cybersecurity professionals and regular users, offering a user-friendly and efficient way to access and analyze vulnerability information. This thesis discusses the limitations of the current implementation and outlines potential future directions for enhancing the chatbot’s capabilities, expanding its data sources, and incorporating more advanced NLP techniques. Also this study will attempt to compare this domainspecific chatbot with other popular general-purposed ones like OpenAI’s ChatGPT and Google’s Gemini.