He will graduate with his Master’s in Information Technology Cum Laude at Nelson Mandela University’s graduation on 14 December.
This is a very relevant problem, and the issue can be seen today with the high volume of COVID19-related misinformation, says Okuhle.
“As for the underlying machine and deep learning technology, this is currently a hot topic and I would definitely use it for business problems that require some sort of prediction capabilities, he says.
Combatting fake news is a complex task and given the immense scale and speed at which online stories are produced, manually fact-checking each one is not possible and therefore automated solutions are needed.
Okuhle’s study investigates the effectiveness of machine learning and deep learning models in detecting text-based fake news. The insights and results obtained are then synthesised into practical guidelines for developing machine learning models that accurately detect fake news.
A series of machine learning algorithms, and several deep learning algorithms were explored in the experiments. In his research to determine whether articles are fake or real, he included text readability metrics such as Flesch Kincaid Grade Level Index, Automated Readability Index and the Gunning Fog Index to calculate the readability of a piece of text.
He also included Cosine Similarity as a metric to calculate the relationship between an article title and article contents as fake news typically have headlines that don’t correlate to the article body. These were the many metrics he used for his machine and deep learning models to detect fake news.
His supervisor was Prof Bertram Haskins, who was “amazing” he says.
His work at Property24 involves developing, maintaining and updating software, which often involves innovating, and solving interesting business problems. This mostly involves back-end (web application, web services, database work, etc) and front-end work. He also does research, always learning new approaches when needed.
He says he can see himself using his master’s research if he ever decides to build misinformation detection systems.
During the course of his study, some of his experimental data was published in a conference paper at the 2020 IEEE Asia-Pacific Conference on Computer Science and Data Engineering (CSDE), titled “Fake News Detection Using Content-Based Features and Machine Learning” (https://ieeexplore.ieee.org/document/9411638). The conference was hosted virtually in Australia.
Okuhle was born and grew up in East London, where he matriculated at Hudson Park High School in 2012. He then went to Mandela Uni and obtained his National Diploma and BTech both in Information Technology.