Exploring AI applications to further business intelligence

Mariam Walaa was inspired to do her Master’s degree in mathematics at TMU because she wanted to work with Professor Anthony Bonato, an expert in graph theory (the study of graphs or mathematical structures used to model relationships between objects). Now in her second year, she is working as a graduate research intern at global consumer goods company Unilever under a Mitacs funding partnership focused on artificial intelligence business applications. 

These experiences have only strengthened Walaa’s interest in contributing to industry when she graduates, to apply her mathematical training to real-world business problems. 

A math major as an undergraduate, Walaa discovered network analysis in a third-year course about graph theory and then deepened her knowledge in a summer course with Bonato. In her thesis project, she applied graph theory to competition networks, looking at open-source data sets available online and investigating their common-out-neighbour or “CON” score to quantify shared influence. While the work expanded on a project Bonato had started with a previous graduate student, this approach had not yet been used. 

In her research, Walaa applied the CON score approach to competitions, including esports, the television show Survivor, and the gaming platform Chess.com. “We tried to figure out if the ‘centrality measure’ that Professor Bonato and his previous student invented could be used as a predictor of rankings of players in different competitions,” explains Walaa, noting that the centrality measure shows how much a player has in common in terms of victories with the rest of the network. It did: in a paper (external link)  that Bonato and Walaa published to arXiv, an open-source research-sharing platform, and will present this summer at a conference in Lithuania, the results show that the CON score sometimes outperformed other traditional centrality measures.

Walaa’s work at Unilever had a completely different focus, aiming to build a model that could improve the way Unilever manages its product portfolio. While Unilever already utilizes AI algorithms and data analysis models, Walaa’s work is experimenting with graph theory to enhance the analysis of the company’s existing consumer data, aiming to make it more predictive. “If they have a portfolio of 10 products, by having this information, you could not only know what previously worked well, but also make predictions on what could work well in the future, without the data being required,” explains Walaa.

Walaa uses popular television competitions such as Survivor and the gaming platform Chess.com as the basis for her exploration into competition networks.