AI research drives better auto sales

Partnership provides big wins for e-company and research team

A Canadian e-commerce marketplace that brings together consumers and car dealerships from across the country is relying on UVic expertise in artificial intelligence (AI) to improve its operations and enhance its customers’ online experience.

VINN Auto offers both a place for paying dealerships to list their new and used cars for sale, and free, personalized support to consumers looking to purchase their next vehicle. 

“The company wants to incorporate effective artificial intelligence tools into its operations—and that’s a very good match with what our research group focuses on,” says Kin Fun Li, a computer engineering researcher at UVic. For about two years, Li and his team of engineering graduate students have worked on a series of projects, identifying areas for improvement and providing recommendations and tools to VINN. 

“At the same time, VINN is enabling our team members to work on real industry problems, not just from a textbook or pure research,” Li adds. “So it’s a two-way street—we’re both learning from each other.” 

Li and the Victoria-based company were introduced in 2019 through Mitacs, a non-profit organization that builds partnerships between academia and industry to support Canadian innovation and foster a highly skilled workforce. The collaboration has been so successful that the first project led to a second, which in turn is leading to a third and a related research paper.

For the initial project, which began in 2020, PhD candidate Linlin Zhang was tasked with cleaning up and restructuring VINN’s database so that when customers enter their information on the company’s website, it’s organized and stored more efficiently.

In the second project, PhD candidate Jing Ning is using AI to help the company’s platform do a better job of classifying consumers who enter their information online. For example, is this a real person or a fake user? Is their email address real? Do they live in North America? Knowing this type of information before a VINN employee contacts a prospective customer saves the company time and money and improves the customer’s experience.

Getting a good PhD or master’s project in artificial intelligence can be difficult, especially getting real data from industry. But these students are doing projects related to their research areas—in this case, artificial intelligence and machine learning—and they’re getting paid to do it.”

—Kin Fun Li, UVic electrical and computer engineering professor

In another project, a master’s student used VINN data as the basis for her research paper, which was recently accepted by a high-profile conference in Japan. Somayeh Abniki’s paper details how to use image processing and deep learning to identify the precise vehicle colour a particular client wants. Abniki now works part-time at VINN (the name is an acronym for “vehicle identification number”), which counts several UVic alumni among its employees.

Tom Avant, co-founder and chief technology officer at VINN, says working with the UVic team has been a great experience.

“VINN is getting this fantastic opportunity to work with talented students, gaining access to academic level machine learning sophistication, while at the same time being able to work within our community and help students get prepared for real-world problems,” says Avant. “It’s a real win-win and allows us to deliver better and better products to our customers.”

Currently, the partners are applying to Mitacs to fund another project that will use AI to interpret and categorize incoming voice messages to ensure company employees can offer the best possible customer service. 

“I think VINN’s competitive edge is in offering free one-stop shopping and personalized customer service,” says Li. “Our research projects are trying to make that experience even better.”

• Read a recent Q&A with Chet Flanagan, a UVic alum and VINN’s chief operations officer.