June 6, 2022

Game-changer for UCalgary researchers in data analytics

New powerhouse server provides additional resources and significantly reduces processing times for complex data
Game-changer for UCalgary researchers in Data analytics

There’s an increased demand for projects leveraging data analytics in the public policy space. Datafication now permeates every aspect of public service work in international organizations, national and regional governments, and even in municipalities. From collecting data, protecting citizens’ privacy while promoting accountability and transparency, providing data-focused tools for decision-making and policy evaluation, encouraging better, adaptive, and personalized programming to facilitate government-citizen relations, we are progressively noticing the automation of public policy.

As big data is becoming ever increasingly central to public policy, the School of Public Policy (SPP) has started a long process of building its expertise in data analytics, developing courses focusing on data analytics and public policy, and building its computational resources. 

Dr. Jean-Christophe Boucher, PhD, assistant professor at the SPP and the Department of Political Science, recognized the importance of using machine learning methods (a form of artificial intelligence that makes predictions from data) in his research early on and has applied that approach to a variety of his projects. One of his specialties is applied machine learning to text analysis, using a combination of supervised and unsupervised deep-learning models in natural language processing. 

Growing presence in big-data world

“Evidence-based decision-making is taken to a new level by data analytics. J-C Boucher’s research team and our new powerhouse server give our School a growing presence in the world of big data,” says Peter MacKinnon, interim director, the School of Public Policy

The international research group led by Boucher has grown exponentially this past year. Boucher’s work has attracted interest from several government agencies. His research is being funded by the Department of National Defence  to study misinformation; the Social Science and Humanities Research Council (SSHRC) to understand civil-military relations in Canada; Alberta Health Services; and grants from Alberta Innovates, the Vaccine Confidence Fund, and Merck to study vaccine hesitancy on social media. 

However, processing very large amounts of data often takes considerable time. Initially, the team was using advanced research computing (ARC) at UCalgary, a high-performance computing resource available to UCalgary researchers. ARC is shared with other departments, meaning longer wait times, and only one job can run at a time for 24 hours due to ARC’s time limitations.

Processing complex algorithms made easy

With the growing needs of his team, Boucher and Abdel Yousif, director of research computing services had to find a solution. In November 2021, a decision was made to purchase a separate partition (nicknamed KABY), which would allow the team to run several jobs simultaneously with no limitations. 

“Processing complex, graph-based algorithms with large sets of data requires extensive computer and memory resources. With millions of graph vertices and edges in the research that Dr. Boucher’s team is performing, there was a need to build a custom server to meet the processing demand of such complex problems.

"The new server (with 8TB of RAM) will help accelerate the processing of graph algorithms by keeping so much data close to the CPU which in turn reduces execution time significantly,” says Abdel Yousif, director, research computing services, Information Technologies, UCalgary

This purchase is significant to the University of Calgary as it allows the SPP team to perform large computations uninterrupted. When not in use, KABY can be made available to other research groups at the university, providing additional resources for their computational needs.

Having a large team of specialized data analysts, software engineers and policy researchers was one part of the equation. Adding a super powerful machine, which performs significantly faster, makes the whole team stronger and more efficient. These improvements increase the likelihood of attracting more grant funding for the study of misinformation and its effects on national defence and security.