A Dataset is a namespace or bucket for collecting or grouping Concepts together.
Data Graphs' project homepage shows all the project's Datasets as cards where you can see the number and types of Concepts each Dataset contains.
Datasets provide two main functions:
1. Information management - a way to efficiently or conveniently manage data
2. Security - a way to make some of your data public or privately accessible to applications at integration time
A Dataset does not add any semantic value to your knowledge graph, but does provide a nice way for you to group data together to improve its manageability.
For example, you may wish to put all Concepts of one type into its own Dataset, such that a Dataset of Organizations would all be collected together. Alternatively you may have defined a Person type and create one Dataset to hold all politicians and another Dataset to hold all sports people.
* Remember - you can still federate across all your Datasets. Putting the same type of Concepts into multiple datasets does not prevent you from searching and navigating data by type across your entire knowledge graph.
A Dataset can be declared as public or private. This defines how the data within the Dataset is accessed by the Data Graphs API.
Private - A Dataset declared as private requires the use of OpenID credentials. This means your data in this Dataset is secure and can only be accessed via an authenticated user request. Machine-2-Machine OpenID / OAuth credentials can be setup in the Application section of the project management settings.
Public - A Dataset declared as public can be accessed using just your API key. These data requests can be made without authentication.