Present Time

Colorful shopping bags surrounding a woman doing online shopping.

Bought a onesie for my friend’s newborn, and now EVERY ad I get is baby-centric! 

This is Sandra Tsing Loh with the Loh Down on Science, saying…

Online shopping algorithms sure learn fast! Hey- YOU MIGHT ALSO LIKE diapers, baby wipes, a nursing bra … What? When you make seasonal or one-time purchase, bots can get messed up. Is there a way to keep their suggestions on point?    

Jianling Wang and James Caverlee from Texas A&M collaborated with e-commerce website Etsy. Together, they created a new online shopping recommendation system. It factors in your purchase history AND seasonal trends. 

To test the system, the researchers used years of anonymous shopping records from Etsy and Amazon. They split the data into two parts. The first was used to train the model, the second — to test its performance. 

How did it do? Better than seven other state-of-the-art models! Even if you buy lipstick and eyeliner frequently, the system STILL remembers you bought a leather bag for dad’s birthday. And will recommend matching gloves next year!

Whoa! Time to buy my roomba his birthday bristles!

Reference: Wang, J., Louca, R., Hu, D., Cellier, C., Caverlee, J., & Hong, L. (2020). Time to Shop for Valentine’s Day. Proceedings of the 13th International Conference on Web Search and Data Mining, 645–653.