Follow Datanami:
October 18, 2019

eBay Challenges Students to Improve Product Identification

Trying to find an exact product online can be a hit and miss affair. Sometimes your search nails it, but other times it misses the mark. Now eBay is enlisting college students to try their hand at developing a machine learning solution to this industry problem.

This week the ecommerce giant launched eBay University Machine Learning Competition, which pits 40 students from four universities against each other with the goal of creating a machine learning model that can accurately identify if two products are the same, given a listing of the product details. The industry calls this Product Level Equivalency.

eBay gives us the lowdown on the importance of PLE. It explains:

“If a buyer purchased two items from two different listings in a single group, and assuming the items were in the same condition, they would assess that they had obtained two instances of the same product,” the company says. “PLE is defined over manufacturer specifications. That is, offer specific details such as condition, price, shipping cost, return details, location are to be ignored. For example, a broken phone and a new phone with the exact same specifications (make, model, color, memory size, etc.) are considered to be Product Level Equivalent, while a golden and a gray phone of otherwise the same make and model are not considered Product Level Equivalent.”

The students will get access to more than 1 million raw (unlabeled) listings from actual eBay sellers. eBay says about 25,000 of those listings will be clustered using human judgment (“true clustering”). These clustered listings will be split into three groups, including a validation set composed of about 12,500 listings, a quiz set composed of about 6,250 listings, and a final submission set that also has about 6,250 listings.


The teams from NYU, Stanford, University at Buffalo, and the University of Texas at Dallas have already begun work on the problem. The competition will run through next year, and big prize is the offer of an internship at eBay.

Senthil Padmanabhan, a vice president and a technical fellow at eBay, describes the significance of the PLE problem in an October 16 blog post.

“At eBay, we use state-of-the-art machine learning (ML), statistical modeling and inference, knowledge graphs, and other advanced technologies to solve business problems associated with massive amounts of data, much of which enters our system unstructured, incomplete, and sometimes incorrect,” he writes.

“The use cases include query expansion and ranking, image recognition, recommendations, price guidance, fraud detection, machine translation, and more,” he continues. “Though most of the above use cases are common among other technology companies, there is a very distinctive and unique challenge that pertains only to eBay – making sense of more than 1.3 billion listings, of which many are unstructured.”

By sharing this unique PLE challenge with the outside world, eBay hopes to grow its community, spur more research in this sector of ecommerce, and improve the eBay platform, he says.

Related Items:

How eBay Leverages Data Visualization in Machine Translation

Hadoop Gets a Security Boost from eBay’s Eagle

‘Tis the Season to Hunt Fraudsters (with Big Data)