Data discovery is a term probably most mentioned in relation to business intelligence and data science. I this context data discovery can be seen as a more experimental and preliminary activity that can lead to a more continuous and integrated form of reporting and predictive analysis when hidden data sources, relationships and patterns are identified.
However, data discovery is useful in other data management disciplines as well.
With the increasing awareness of data security, data protection and data privacy – and the regularity compliance enforced in this space – it is crucial for organisations to know what kind of data that flows and are stored within the organization. While you may argue that this should be available in already existing documentation, I have yet to meet an organization, where this is the case. And I come around a lot.
Data discovery is also a component of test data management and tool vendors package their offerings in this space with capabilities for data masking, data subsetting and data discovery in order to answer questions as:
- Where are the data elements that should be masked when using production data in test scenarios without violating data privacy regulations?
- How can you subset (minimize) test data sets derived from production (covering several databases) and still have proper relationships covered?
Within Data Quality Management, Data Governance and Master Data Management (MDM) data discovery also plays a role similar to the role in data reporting. We can use data discovery to map data lineage, find potential data relationships where data matching, data cleansing and/or data stewardship might help with ensuring data quality and business process improvement and explore where the same data have different labels (metadata) attached or the same labels are used for different data types.