Determining complementarity between products when viewing products online, e.g., via a shopping website or a search engine, is not straightforward. The information required to determine product complementarity is unstructured and/or spread across multiple sources. While co-purchase data can be used to infer complementarity, this requires large volumes of purchase data that is often unavailable. This disclosure describes techniques to identify relationships between products based on analysis of web search queries, subsequent clicks or other user actions, obtained with user permission. The techniques rely on the observation that query streams often include searches for related products, e.g., in the form of “A for B” queries (or queries in other forms) when searching for a product A compatible with a product B. Entity extraction is performed on such query streams and a machine learning model is utilized to identify likely complementary products.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Benavides, Luis; Shur, Timothy; Sawant, Anshul; Gollapudi, Sreenivas; Parikh, Nishith; Gupta, Akshay; and Xu, Leon, "Determining Product Complementarity By Analyzing Search Query Streams", Technical Disclosure Commons, (March 18, 2021)