All concepts that describe a particular domain (such as effective exchange rates of the euro) are grouped into a data structure definition (DSD).

Time series with related content are collected under a DSI using the same key structure. The key structure is defined in a DSD. Each DSI is assigned to exactly one DSD.

An exception to this rule is the DSI “BBK01”, which is not assigned to any DSD.

Several DSIs can be assigned to one data structure definition (e.g. for a current and a historical dataset).

There are two concept types – dimensions and attributes.

When combined, dimensions allow statistical data to be uniquely identified. Attributes provide additional useful information (such as source of data).

Attributes can be free texts (e.g. a comment on a particular observation value), dimensions may only take on values from a predefined list (such as a list of countries) or must follow a pattern in the form of a regular expression (e.g. [a-zA- Z0-9] {1.5}).

Predefined lists are called code lists. They contain all versions of a key component of a time series key (e.g. the country code list contains the codes DE, FR, IT, ...). Code lists and patterns can also be used for attributes.

Dataflows represent the data that relate to a specific domain (e.g. exchange rates, euro reference rates). A dataflow indicates the data structure definition that applies to a particular domain and thus specifies how the data for that domain need to be structured.

A time series key usually consists of several components, the first of which determines the frequency. The keys in the BBK01 dataflow are the exception to this rule. They consist of between six and ten digits, without a separator between the components. The second digit indicates the frequency.