GitHub user clhubert created a discussion: Build Trust in Data and Enhance 
Semantic Layer for AI via Data Provenance

BRIEFLY Discussed this with @rusackas / @michael-s-molina  in the 7/1/2026 
Super Set SIP Review.
Looking for feedback on usefulness and potentially clarity on an a future 
extension/SIP.

**Use Case 1** :  Build Trust in Data AND Enhance Semantic Layer for AI via 
Data Provenance 
- The source of the data in a superset dataset can be used to ENHANCE AI's 
semantic awareness via its ability to trace the linage AND provenance of data. 
- For the report/dashboard end user who want to know WHERE the data come from, 
like a reference, footnote, etc...   Build trust in data provided in dashboard.

**Use Case 2** - Visualize:  Report Author/Viewer wants to see 
"**_provenance_**" of data from a dataset visually like a sql trace. 
- Linage is already supported with what’s available today: community extension. 
By @michael-s-molina 
- https://superset.apache.org/developer-docs/extensions/registry/, 
https://github.com/msyavuz/superset-sql-visualizer
- Including data provenance leads to more COT (Chain of Thought), for the AI.

**Use Case 3:** Pull in EXTENDED Data Linage/Provenance MetaData
- This is beyond where the information is at rest in data set, but where the 
data that was inserted into the dataset came from (Provenance).
- This data may already be accessible via a Data Catalog, Glossary, MDM and or 
Data Governance Platform such as Open Meta Data 
- Open MetaData - pulls in metadata from multiple sources   
-- Data Dictionary terms in glossary can be associated with several fields.  In 
OpenMetaData this would be associated an asset(s) to a term. - 
https://docs.open-metadata.org/v1.12.x/how-to-guides/data-governance/glossary/glossary-term
 .   Other MDMs and Catalogs (Talend Data Catalog), have something similar. 
--- OpenMetaData uses Apache Airflow for Ingestion - 
https://docs.open-metadata.org/v1.12.x/connectors/pipeline/airflow
--- **Lineage Workflow** - 
https://docs.open-metadata.org/v1.12.x/how-to-guides/data-lineage/workflow
--- **Lineage Ingestion** - 
https://docs.open-metadata.org/v1.12.x/connectors/ingestion/lineage
--- Apache Nifi is one of the Source - 
https://docs.open-metadata.org/v1.12.x/connectors/pipeline/nifi, which exposes 
its internal data provenance information to OpenMetaData.

**Notable:**
- Not clear if the Data Governance use case mentioned above can be accomplished 
using an MCP enabled Data Governance Platform.
-- Open MetaData has MCP integration into their semantic layer. 
https://open-metadata.org/mcp
- Avoid building a federated query engine - Denodo, TIbco Data Virtualization 
that does query plans across disparate data sources. 

GitHub link: https://github.com/apache/superset/discussions/41633

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: 
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to