eBay Taps Students To Solve Fitment Challenge In 2024 University Machine Learning Competition
eBay's annual university competition tasks US students with solving vehicle fitment challenges with winners vying for a spot in the 2025 eBay summer internship program.
eBay serves up a unique ecommerce challenge and dataset for each year’s contest, inviting qualified students to team up and find innovative solutions to the problem at hand. This time around, participants will delve into the world of eBay Motors. Their mission: Build a model that can identify which vehicles are compatible with specific parts and accessories.
This compatibility – also known as “fitment” – is a crucial factor for sellers and buyers alike. Extracting fitment data from listing titles, item specifics, and descriptions within the eBay Parts and Accessories categories is a delicate task, since the listing data often contains noisy or extraneous information.
Students who enter this year's competition will challenge will be asked to provide solutions for fitment at the “Year Make Model” level using machine learning to extract compatibility information from text and images using sources like listing titles, descriptions, and listing specifics.
The data set will consist of 5 million randomly selected unlabeled items from five parts and accessories category groups, with 1 million items per group. For each of the items the listing title, item specifics, and description will be provided, along with a category group ID.
Among these items there will be 10,000 items where fitment data is known. Fitment data might have been obtained through separate means from the item titles, item specifics, or item descriptions.
The 5 million unlabeled item set, and the fitment training set is intended for participants to build their models/prediction system. The actual fitment data will be provided for each item in the training set, along with the item record id/number to link the item data in the other files.
Sign up is open from May 1st - September 10th, with a deadline for submissions of November 4th and winners will be chosen and notified on November 15th.
Details:
eBay has long struggled with how to handle fitment and compatibility data in the Motors Parts & Accessories categories to make sure buyers are able to find parts that fit their specific vehicles, acquiring MyFitment in 2022 to try to tackle this difficult task.
Liz Morton
Clearly that acquisition alone was not enough, as developing reliable fitment solutions still seems to be beyond the reach of eBay's current crop of AI and Machine Learning engineers - thus necessitating bringing in unpaid student labor to help try to solve the problem.
Fitment was seen by eBay as key to unlocking Motors Parts and Accessories as part of CEO Jamie Iannone's vertical focus playbook by introducing a Guaranteed Fit program that promised to provide free returns if an item doesn't fit the buyer's vehicle.
But buyers have been confused and frustrated by the program, being led to believe their purchases will qualify for the guarantee when viewing the item page only to find out they are not eligible when trying to request a return.
Liz Morton
Will college students prevail where eBay has so far stumbled? Stay tuned for results from the winners in November!
