Re-Ranking Layer for Product Search Engine
Walmart.com maintains an online catalog of over 2M products. Consequently, enabling users to quickly find products that conform to their specific needs and tastes is especially challenging. Given the difficulty of its task, Walmart.com’s product search engine does an impressive job of interpretting the user-provided query and rapidly returning relevant results. Yet, there remains highly significant information that is not fully leveraged. The details of a user’s online shopping session are indicative of a user’s intent and compliment --indeed, provide context for—the user-provided query. This code is the implementation of an ordering scheme we call Session Re-Rank (SRR) that can potentially induce a large increase in both click-through-rates and conversions on the first page of query results.