Dr. Shavindrie Cooray Publishes and Presents New Research on Bias in Machine Learning Data Analysis and Social Network Influence

This summer, Dr. Shavindrie Cooray, associate professor in management information systems, published her research in the Information Society journal (Taylor and Francis Group publishing) which has an acceptance rate of 5 to 6 percent and a 2020 impact factor of 4.57.

Her sole-authored research seeks to demonstrate how an analyst’s subjectivity can have a bearing on both qualitative and quantitative social network analytics. The paper also employs quantitative and qualitative techniques to determine those who are most influential in social media groups and gather insight about what resources, characteristics, and types of knowledge helped them gain influence.

Dr. Cooray also was recently invited to publish an article, ‘Analyzing the Analyst: A Guide to Holistic Analytics for Tracking the Right Metrics,’ on the MarketingProfs website, which has more than 650,000 subscribers and a similar number of followers on social media.

She was selected to present her research this summer at the 26th Americas Conference on Information Systems, 39th International Systems Dynamics Conference and the 21st Conference of the UK Systems Society.

Currently, Dr. Cooray’s research focus is on analyzing bias in machine learning/AI. She points out that “there's a misconception that in using artificial intelligence and machine learning, people are removed from the experience and only machines make predictions.

“That's completely incorrect because it's the people who decide which data sets to use for training and what is going to be contained in those data sets,” she says.

Dr. Cooray further notes that “the predictions made by the machine will only be as good as the data that it is trained on. The data sets are designed by people. That’s why it’s important for students to understand these new concepts.

“I’m really passionate about machine learning and artificial intelligence, especially with more companies using technologies such as AI and machine learning in their decision-making process,” she adds. “It’s important for us to consider whether those technologies are actually fair and less biased toward often-marginalized populations.”

Dr. Cooray’s current research on the use of systems thinking/systems dynamics to reduce bias in machine learning datasets is currently under journal review. She points out that her research influences her teaching, as she introduces new concepts to her students in the classroom in order to give them the tools needed for success in today’s workforce, such as in her MIS 2050 Data Mining, Artificial Intelligence and Machine Learning course being taught this fall.

It’s a required course for both the Management Information Systems (MIS) minor and the newly-launched Data Analytics minor. Concepts such as data management, artificial intelligence and machine learning are covered in both the MIS and Data Analytics minors with the Data Analytics minor having a more mathematical focus and the MIS minor a more business focus. The newly-launched marketing major is also data-driven and includes classes such as Marketing Analytics, Digital Marketing and Marketing Research.

“It’s a result of demand because now it's all about data analytics and machine learning is a big part of that,” Cooray points out. “We try to bring in new programs and build relationships as part of these programs with businesses like DataRobot, whose software students will be using in my class.”