Cyber patrol threat detection through a multi-lingual AI algorithm

Author

Mukaro, Justice

Title

Cyber patrol threat detection through a multi-lingual AI algorithm

Abstract

Billions of tweets are sent out daily in all sectors of life especially in business and governance in multiple languages. Some of the tweets sent out present threats to leaders and also influencers utilise the platform to incite violence and cause civil unrest. This research brings about the creation of a multilingual threat detection algorithm to scour through selected tweets and conduct a threat analysis. Using Twitter APIs, the bot legally extracts tweets and uses the Python language translation function and then it analyses them using Natural Language Processing (NLP). The data is collected, cleaned, analysed and output is given on a word cloud and other statistical representations. The algorithm was created on the Google Collaboratory platform which is a browser-based Python compiler. It uses Google Servers to access RAM, GPU and competent processors. Thus, all the output is automatically stored on a Google Drive. The researcher collected data from the President of the United States, Joe Biden. His tweets were analysed for threats and or abuse of power and also, tweets directed towards were analysed for the same information. The output was presented on a Word Cloud, scatter diagram and a bar graph.

Date

JUNE 2024

Publisher

BUSE

Keywords

multilingual threat detection algorithm,

Using Twitter APIs,

Natural Language Processing (NLP),

Supervisor

Mr. Zano,