The schema of what type of data can be defined at query time?

The schema of unstructured data can be defined at query time because unstructured data lacks a predefined data model, allowing for a flexible approach to organization and analysis. This means that users can specify how they want to interpret or analyze the data when the query is executed, adapting to various needs without the constraints of a fixed schema.

In contrast, structured data adheres to a strict schema defined beforehand, making it less flexible in terms of how queries are formulated. Azure Cosmos DB is a multi-model database service that can support both structured and semi-structured data but does not inherently allow for schema definition at query time in the same manner as unstructured data. Time-series data typically has a defined structure over time to accurately represent and analyze temporal changes, making it similar to structured data in its schema rigidity.

Thus, the nature of unstructured data allows for this dynamic schema definition at the time of query, making it the correct choice.