Exploration of Product Search Intents via Clustering of Query Clusters

We describe a system that organizes search results in the context of an exploratory product search session where the user is researching goods. Compared to existing approaches that use predefined categories to filter results by attributes, we organize information needs based on queries instead of documents. The idea is to organize queries around the same topic and produce a hierarchical representation of intents that describe information about a product from different perspectives. We present a prototype implementation using a real-world data set of 24M queries.