Data-level security defines what rows or columns are returned to a user inside an analysis or DV project view. The same view, when run by two different users, can bring up different data. The structure of the view is the same for all users unless a user does not have access to the subject area, in which case the report displays an error.
You can read about row-level or data level security topics in various places in the documentation and it may depend upon your approach (data modeler, RPD, or data set)
For a primer, refer to About Data Level Security
The recommended approach is to use the Enterprise Semantic model for data-level security (row-level) to be applied to a data visualization (dv) project. The project must use an RPD subject area as the data set. This is the most robust solution for enterprise and departmental use cases.
You can read a comprehensive step-by-step Oracle blog post: Implementing Data Level Security in Oracle Analytics Cloud Using Identity Cloud Service
Note: For this use-case, data visualization projects must use a metadata repository (RPD) subject area for the data set.
