Navigation using metadata exposed in a semantic zooming environment. The semantic zooming environment has multiple zoom levels. For a given displayed object, the displayed object may not just be a different size, but actually be displayed differently at each semantic zoom level. At one or more of the semantic zoom levels, a displayed object may have displayed metadata of the displayed object. A user may interact with the metadata to trigger a query on that metadata against a data source, causing an entire collection of objects to be populated that also have some relationship to that same metadata. The collection of objects may itself be populated within a semantic zoom environment that additionally exposes metadata, allowing the process to repeat. Thus, the principles described herein allow advanced navigation techniques whereby data may be narrowed down and/or change direction, easily by users manipulating metadata exposed in semantic zoom environments.