Making Sense of Search: Using Graph Embedding and Visualization to Transform Query Understanding

We present a suite of interfaces for the visual exploration and evaluation of graph embeddings -- machine learning models that reveal implicit relationships not directly observed in the input graph. Our focus is on the embedding of navigation graphs induced from search engine query logs, and how visualization of similar queries across different embeddings, combined with the interactive tuning of results through multi-attribute ranking and post-filtering (e.g., using raw query frequency or derived entity type), can provide a universal foundation for query recommendation. We describe the process of technology transfer from our applied research team to the Microsoft Bing product team, examining the critical role that visualization played in their decisions to ship the technology on bing.com.