Julius Onyancha has beat hundreds of other entrants to win The Best Student Paper Award at this year’s International Conference of Data Mining and Knowledge Engineering, organised by the World Congress of Engineering in London.
The news was revealed during the international Computer Science Education Week (Dec 4-10) an annual programme dedicated to inspiring students to take interest in computer science.
A PhD student in the University of Sunderland’s Faculty of Computer Science, Julius impressed judges with his paper ‘Learning from Noise Web Data’, which looks at developing tools to decrease levels of irrelevant and meaningless ‘noise data’ as users click through websites online while at the same time preventing the loss of useful information.
Kenyan-born Julius said, “Given the number of paper submissions to the conference, I was just happy to have my paper accepted for publication and presentation. What I did not expect is selection and award for best student paper. I appreciate the recognition and must admit it is a good opportunity to demonstrate my research contribution to the data science community. I received a lot of feedback from the participants at the conference who were keen to find out more about my research.”
The winning paper was selected based on reviewer reports and the evaluation score results of committee members during the conference.
Explaining the concept behind his winning paper, Julius said: “Due to the increase in data available on the web, it is becoming difficult to find useful information without encountering irrelevant or meaningless noise data. Web users get frustrated when a website promotes content that isn’t tailored to their interests.
“This paper explores how current available tools address problems with noise in a web user profile. The current research works by eliminating noise from web data mainly based on the structure and layout of web pages. They consider noise as any data that does not form part of the main web page.
“However, not all data that form part of a main web page is of interest. The ability to determine what is noise and useful to a dynamic web user profile has not been fully addressed by current research works. The paper aims to justify a claim that it is important to learn noise prior to elimination, not only to decrease levels of noise but also reduce loss of useful information. This is because if noise in web data is not clearly defined and analysed through learning, the purpose and its use will be compromised as will its overall quality.”
Julius first joined the university in 2007 at an Expo Fair organised by Intel College-Kenya in partnership with University of Sunderland. He enrolled for a degree in Applied Business Computing and later a Masters in Information Technology Management in 2010. He briefly worked for a local IT company supporting various business in building data models for digital marketing campaigns.
He said: “I came to learn that the success of any business organisation is the ability to tell a good story about your customer based on available data. This was the inception of my PhD research work ‘Learning Noise in Web Data’ which I started in 2014.”
Valentina Plekhanova, Senior Lecturer in the university’s Faculty of Computer Science, supervised Julius’ Masters Project where he first used knowledge of machine learning and data mining subjects as part of the degree programme, described him as a “hardworking young researcher”. She adds, “We are very pleased that his research work is supported not just by our colleagues but also internationally recognised as one of the best student’s work.
“A key point of Julius’s research is that we could learn very important and critical information from noise web data. There are many potential practical applications of the research outcomes. It defines a good research niche for Julius’s work.”
Reflecting on his time at university, Julius said: “Sunderland is a life-changing community. As an international student, the university has nurtured my career development and achievement; more particularly the research culture, and ever supportive academic staff who are always approachable and supportive. The faculty supported financially towards publication of this paper.”