Neural Product Retrieval at Walmart.com
Alessandro Magnani, Feng Liu, Min Xie, Somnath Banerjee · 2019
For an E-commerce website like Walmart.com, search is one of the most critical channel for engaging customer. Most existing works on search are composed of two steps, a retrieval step which obtains the candidate set of matching items, and a re-rank step which focuses on fine-tuning the ranking of candidate items. Inspired by latest works in the domain of neural information retrieval (NIR), we discuss in this work our exploration of various product retrieval models which are trained on search log data. We discuss a set of lessons learned in our empirical result section, and these results can be applied to any product search engine which aims at learning a good product retrieval model based on search log data.