Ten years ago, when Illinois computer science professor Hanghang Tong co-published a paper on bipartite graph arrangements, he admitted that he had no way of knowing the future implications of his work.
In his first year out of academia as a researcher at the IBM TJ Watson Research Center, what Tong knew was that a three-month project had proven why he chose a position in the industry. The key findings of this paper proved that his research in data mining could lead to important real-world applications.
Tong worked with intern Danai Koutra and his manager, David Lubensky, on how to apply graph placement to some important social media questions. First, how can the system find a “virtual twin” on another social media site, such as LinkedIn or Facebook? And how can information and social networks be effectively linked to support cross-network searches?
As a result, a paper titled “BIG-ALIGN: Fast Bipartite Graph Alignment” presented “a new optimization formulation and a proposal for an efficient and fast algorithm to solve it”. Their results yielded a method that he found 10 times more accurate and 174 times faster than previous methods.
Presented at the IEEE International Conference on Data Mining (ICDM), the top conference in the field, the work was impactful enough to win an ICDM award. Best Influential Paper of the Decade Late November.
“I was talking to Danai recently and neither of us could believe that 10 years had passed. It felt like it was just yesterday,” Tong said. “At the time, we knew the paper would be of interest among researchers in the field, but winning the Best Influential Paper of the Decade award means a lot to us. That’s it.
“One of the things I can cite from the letter that left an impression on me is that the rationale for this award is based on the committee’s belief that your paper has ‘had the greatest impact on the data mining community in the past decade.’ about it. “
Not only did the paper yield a prestigious and important technical achievement ten years later, it also set new career paths for both Tong and Koutra, the project’s young researchers.
At that point in his career, Tong had a Ph.D. Carnegie He holds a PhD in Machine Learning from Mellon University. student at the time.
Believing that his research in this area could have an impact on academia, Tong showed how directly relevant his research could be to practical applications through his research directly linked to industry. I wanted to check
The papers they produced influenced both e-commerce and the financial sector.
This was quickly appreciated by the review board IBM used to estimate the IP value of the project. In fact, the project received IBM’s highest rating, indicating very high potential business value.
The company’s review board recommended applying for seven patents instead of one.
“I had the benefit of working with Danai, whom I met at CMU. challenged us to explore the most meaningful questions, but he also provided us with a highly nourishing environment that left us free to find our way. said Ton.
“Looking back, what I am really proud of is the impact this project has had on the long-term success of both my career and Danai. , continues to work with students on network coordination, and I have recently completed two PhDs, and my PhD graduate thesis was on this topic on network coordination.”
In total, Tong has spent nearly three years in research at IBM, and credits his success at IBM with helping him find his way back into academia.
He has been teaching at Illinois CS since 2019, but it was this project at Koutra that helped him in part understand how the energy of learning with students impacted his work. It was an experience.
“The experience at IBM was great for me, and it also taught me that I missed the university environment. said.
