Search Results Clustering
Part of the Query Understanding series by Daniel Tunkelang.
Overview
Search results clustering groups the results of a query into thematic categories at display time, giving users an overview of the diversity in the result set and allowing them to navigate to the cluster most relevant to their needs. It is most useful for ambiguous or broad queries where a single ranked list obscures the fact that the results represent multiple distinct interpretations or subtopics. The main practical challenges are generating cluster labels that are accurate and descriptive enough to be useful, maintaining low enough latency that clustering doesn’t degrade the search experience, and designing a presentation that users will actually interact with — clusters that go unnoticed don’t help anyone.
Note: Article content behind Medium paywall — accessible at source URL with Medium account.