Unstructured data stores¶
Unlike structured data, unstructured data does not have a formal or strict "blueprint", often requiring Artificial Intelligence (AI) or Natural Language Processing (NLP) to analyze. Therefore, you can only use unstructured data connections for vector database (VDB) creation. Examples of unstructured data include media files (PNG, MP4) and text data (PDF, text files).
Data store vs. data connection
In DataRobot, data store and data connection are used interchangeably, and both refer to where your data is stored and managed as well as the dynamic link to that data. You will see both terms in the UI.
The following table lists all data stores supported by DataRobot for unstructured data ingest:
| Connection | Connection type | Credential type | Notes |
|---|---|---|---|
| ADLS Gen2 | Connector | OAuth, OAuth service principal | Required parameters and connection details |
| AWS S3 | Connector | AWS credentials (both temporary and not temporary) | Required parameters and connection details |
| Box | Connector | JWT | Required parameters and connection details |
| Confluence | Connector | Basic (username/API token) | Required parameters and connection details |
| Databricks | Connector | Personal access token, service principal | Required parameters and connection details |
| Google Drive | Connector | Google Cloud OAuth, service account | Required parameters and connection details |
| Jira | Connector | Basic (username/API token) | Required parameters and connection details |
| Managed cloud OAuth | Connector | Managed cloud OAuth | OAuth provider management |
| OneDrive | Connector | Azure OAuth, service principal | Required parameters and connection details |
| SharePoint | Connector | Azure OAuth, service principal | Required parameters and connection details |