Building the future of social networks

Sepideh Banihashemi has devoted the last several years to her graduate studies at Toronto Metropolitan University (TMU). She started her journey in 2015 with her master’s in computer science, and, equipped with strong knowledge from her courses and support from the faculty, continued on to pursue a PhD. Last semester, after building upon her lab’s research on social networking technology, she successfully defended her thesis. Now, she reflects on her time at TMU.

What has been your proudest accomplishment at TMU?

My proudest accomplishment at TMU was successfully defending my PhD, during which I played a key role in developing a unique social network simulator. I contributed significantly to its initial development and enhanced it by implementing additional features, improving its functionality for both academic and practical applications. As a member of the Distributed Systems and Multimedia Processing (DSMP) lab, where this software was originally developed and tested, I collaborated with undergraduate, graduate and PhD students, refining the tool through various research projects. My PhD thesis supervisor, Dr. Abdolreza Abhari, the director of the DSMP lab, guided me through a challenging yet rewarding journey. My active involvement in every stage of the DSMP simulator’s development lifecycle, along with my leading role in releasing its third version, provided invaluable experience and growth opportunities. Seeing the DSMP simulation software emerge as a helpful research tool has been a significant highlight of my academic journey.

Your research explores the impact of machine learning algorithms on social networks. Can you break that down for us?

My research examines how machine learning algorithms influence the structure and behaviour of social networks using a simulation-based approach. The simulator serves as a valuable tool to help researchers and educators better understand network dynamics to improve network growth and build stronger communities. The findings of this research show that the number and quality of communities depend on the accuracy of the algorithm, which varies depending on the dataset used. As artificial intelligence and machine learning algorithms continue to improve, along with advancements in parallelization and scalability – that is, the ability to divide and distribute computational tasks into smaller sub-tasks that can be executed simultaneously across multiple processing units – social networks will likely become more interconnected. This means that information will spread more efficiently among users, benefiting applications such as recommendation systems that increasingly rely on machine learning algorithms, social media monitoring, sentiment analysis and influence analysis.

What were some challenges and rewards you faced pursuing a PhD?

One of the challenges I faced while pursuing my PhD was conducting applied research in an area where leading companies with vast resources already dominate the field. Academic research requires innovation and creativity to produce novel contributions, often with limited resources, unlike industry, which benefits from greater resources and support. This meant that my work had to be technically sound and unique enough to stand out and push the boundaries of existing knowledge. To overcome this challenge, I focused on developing a unique social network simulator that allowed researchers to study the impact of machine learning algorithms on social networks. This required integrating advanced machine learning models while ensuring computational efficiency and scalability. Through rigorous experimentation, optimization and collaboration, I built a tool that provided meaningful insights into social network dynamics and their important applications, such as recommender systems.

Despite the challenges, the rewards were significant. Seeing my research evolve into a functional system that could be used for further studies was incredibly fulfilling. Additionally, the process helped me develop a strong problem-solving mindset, advanced my technical expertise, and allowed me to contribute to the broader research community. Successfully defending my PhD and knowing that my work has practical applications in academia and beyond has been one of the most rewarding aspects of this journey.

How has your experience at TMU and the DSMP lab shaped your research journey and prepared you for the next steps in your career path?

Being a member of the DSMP lab has equipped me with a strong foundation for conducting independent research and solving complex problems. The skills I developed in areas like machine learning, social network analysis and software development, along with the collaborative experience in multidisciplinary research, have prepared me to contribute meaningfully to advancing knowledge and innovation in the simulation field. Although simulation is a multidisciplinary field, it has a small community of researchers. I had the opportunity to publish the results of my research (external link)  about the new features of DSMP simulation software in one of its important and impactful venues called Simulation: Transaction of SCS. Being a DSMP lab member has also connected me with the simulation community and allowed me to review state-of-the-art papers about social network simulation. With at least two graduates from our lab in the past three years participating in annual simulation conferences, we have successfully positioned TMU alongside top world-class universities in international conferences on simulation. This recognition will be highly beneficial for my future job search in the fields of simulation and social network analysis.