The central, organizing basis for managing structured data within conStruct is the dataset.

Datasets contain one or more data records from a single source representing the same type of instance(s). Datasets may reside on the Web as well as be stored locally. Each dataset is uniquely identified with standard metadata characterizations.

At minimum, datasets have a simple structure of attribute-value pairs for each instance record. However, they may also have more complex structure via schemas (ontologies) that also describe the relationships between concepts and attributes and may even relate those to external schema.

All conStruct tools operate against one or more datasets, which can be selected for these operations. Individual users may be assigned access rights or not to each of these datasets, and whether they have CRUD (create-read-update-delete) permissions or not.

The combination of access rights and permissions then defines which tools and what operations are available to a given user for each dataset. At present, though being modified, these permissions are set through the Organic Groups module within Drupal.