User roles in Knowledge Studio
IBM Watson® Knowledge Studio roles are used to organize a team of people who create machine learning or rule-based models.
Notes about roles in Knowledge Studio
- IBM Cloud® roles are different from Knowledge Studio roles.
- IBM Cloud® roles control permissions for managing IBM Cloud services. New service instances of Knowledge Studio are managed with IBM Cloud Identity and Access Management (IAM) in all public regions.
- Knowledge Studio roles control access to Knowledge Studio functionality and are managed in the Knowledge Studio application.
- The first user to launch the Knowledge Studio application is assigned the Admin role.
- Once assigned, user roles can't be downgraded from higher to lower levels of permissions. Admins can't be downgraded to project managers or human annotators, and project managers can't be downgraded to human annotators. For information about adding users, upgrading roles, and deactivating user accounts, see Assembling a team.
- To manage a workspace, project managers need to be assigned to the workspace by an admin.
- Admins and project managers can perform the role of human annotators. They can also directly annotate document sets without creating annotation tasks.
Knowledge Studio role descriptions
Knowledge Studio roles are described in the following table. Roles are listed in order of highest to lowest level of permissions.
Role | Description |
---|---|
Admin | Responsible for administrative tasks, which include managing users, resource consumption, and monthly charges. In large team settings, admins rarely participate in the model development process. |
Project manager | Responsible for the overall organization of the workspace that she or he is assigned to. Tasks include creating the type system, managing assets, managing annotation work, evaluating the machine learning model, and deploying models. Users in this role need industry subject-matter expertise because they create the type system, teach the human annotators how to correctly apply the type system, and evaluate the model quality. |
Human annotator | Performs the labeling of the entity mentions and relationship mentions in the training documents that he or she is assigned to. The work is assigned to human annotators and monitored by the project manager. Human annotators may not have industry subject-matter expertise, as long as they are taught by the project manager how to correctly apply the type system. |
Knowledge Studio role permissions
To compare the permissions of each role, see the following table. One permission is listed on each row. If a role has that permission, the role column is marked with a check mark (✓).
Permission | Admin | Project manager | Human annotator |
---|---|---|---|
Manage user access and roles | ✓ | ||
Manage workspaces | ✓ | ||
Create the type system and annotation guidelines | ✓ | ✓ | |
Upload training documents | ✓ | ✓ | |
Manage rules | ✓ | ✓ | |
Pre-annotate documents | ✓ | ✓ | |
Manage annotation tasks | ✓ | ✓ | |
Resolve annotation conflicts with human annotation (adjudication) | ✓ | ✓ | |
Train and evaluate machine learning models | ✓ | ✓ | |
Deploy and undeploy models to runtime services | ✓ | ✓ | |
Annotate document sets directly | ✓ | ✓ | |
Perform document annotation in annotation tasks | ✓ | ✓ | ✓ |