Skip to main content

I didn’t even know I needed that!

eBay Buys Hunch To Improve Long-Tail Shopping Recommendations


Online shopping has quickly replaced commercial businesses for consumers. Companies such as eBay, amazon, google, etc use complex algorithms and optimization methods to best recommend products based on individuals previous searches. Recently eBay hired a company, Hunch, that is a leading company in data mining, machine learning, and predictive performance. The co-founder of Hunch mentions that their integration with eBay will not only allow eBay to generate more revenue and sales, but also will help customers find meaningful and useful purchases of things they haven’t thought of before. For example, on many sites there is a feature in which they recommend “users who bought X, also bought Y.” The company Hunch will be working to optimize the algorithm for those suggested purchases section.

When using topics in class to understand how a company like hunch is making over $20M in sale for a team of 20, you identify the root of the algorithm to use the triadic closure property. It finds that you and a user have a very strong tie to one product and forms another tie between you and that user. And then generates more ties between you, the item, and other uses to create a weighted network on strong and weak ties and uses those weights to calculate the best recommended product for your purchase.


What I find so interesting is their ability to collect and classify enough data to determine the weight of the tie. There are so many factors to account for in these algorithms and the strength of the ties is a factor of profile, interests, mutual towns, age, gender, etc. I believe that it would be fascinating to see all the inputs of the predictive model and see how the inputs vary based on different users. Reading this article shed light to the fact that networks has a far larger application to data driven platforms and their effects on society. Essentially data driven platforms are the only tools and methods we have for predicting our future.


So next time you see the “recommended for you.” Consider it strongly because who know when in the future you might be saying, I didn’t know I needed that!


Leave a Reply

Blogging Calendar

September 2018