Open Access

ARTICLE

AI-Enabled Grouping Bridgehead to Secure Penetration Topics of Metaverse

Woo Hyun Park¹, Isma Farah Siddiqui³, Nawab Muhammad Faseeh Qureshi^2,*

1 Department of Electrical and Computer Engineering, Sungkyunkwan University, Suwon, 16419, Korea
2 Department of Computer Education, Sungkyunkwan University, Seoul, 03063, Korea
3 Department of Software Engineering, Mehran University of Engineering & Technology, Jamshoro, Pakistan

* Corresponding Author: Nawab Muhammad Faseeh Qureshi. Email:

Computers, Materials & Continua 2022, 73(3), 5609-5624. https://doi.org/10.32604/cmc.2022.030235

Received 22 March 2022; Accepted 17 May 2022; Issue published 28 July 2022

Abstract

With the advent of the big data era, security issues in the context of artificial intelligence (AI) and data analysis are attracting research attention. In the metaverse, which will become a virtual asset in the future, users’ communication, movement with characters, text elements, etc., are required to integrate the real and virtual. However, they can be exposed to threats. Particularly, various hacker threats exist. For example, users’ assets are exposed through notices and mail alerts regularly sent to users by operators. In the future, hacker threats will increase mainly due to naturally anonymous texts. Therefore, it is necessary to use the natural language processing technology of artificial intelligence, especially term frequency-inverse document frequency, word2vec, gated recurrent unit, recurrent neural network, and long-short term memory. Additionally, several application versions are used. Currently, research on tasks and performance for algorithm application is underway. We propose a grouping algorithm that focuses on securing various bridgehead strategies to secure topics for security and safety within the metaverse. The algorithm comprises three modules: extracting topics from attacks, managing dimensions, and performing grouping. Consequently, we create 24 topic-based models. Assuming normal and spam mail attacks to verify our algorithm, the accuracy of the previous application version was increased by ∼0.4%–1.5%.

---

Keywords

---