Activate 2018

To further improve discoverability of Macy’s product catalog online we introduced an easy shopping experience for finding products which are hard to describe using text-based search. The feature allows customers to use an existing product on the website and find all matching products which are similar to the original product based on its image. Deep-learning algorithms are applied to product images to provide an experience. To enhance it further a new component is developed to evaluate the signals other that image similarity, such as product attribute similarity, customer’s shopping preferences, business rules etc. Current talk gives some insights on implementation and overall feature architecture.