DAMA Session - Ensuring Trusted Data Through Managed Import and Semantic Integration 

 Ensuring Trusted Data
Through Managed Import and Semantic Data Integration

Presented by Michael Scofield    
 More enterprises are importing data from external sources (which they cannot control), and integrating that data either into their operational systems or into their data warehouse.   Data from external (uncontrolled) sources can be problematic because (1) their
logical data architecture can be different, and (2) they can morph their
architecture, scope, or definitions of the data which you receive without telling you.  Hence, the need for a rigorous declaration of expectations in the agreement to establish that data flow and business relationship.  

           If the imported data is used for operational systems (for example, expanding the knowledge about a customer),then this is enhancing a portion of the “master data” of your enterprise.
 A major challenge is proper data integration.  Their ID’s for
customers may not match your ID’s.  And their coding scheme for relatively
stable customer (or product) attributes may differ.  Hence, a daunting
effort to design a strategy of data import, screening, translation, and
integration.   

             One important goal of master data
management is ensuring that changes (internal or externally driven) do not
cause harm to downstream systems.  The most significant downstream system
sometimes is the data warehouse driving decision-support and business
intelligence.  One major technique to enrich data points to convert them
to usable information is to place each data point in context--context of
history or context of peer enterprises (e.g. competitors).  History
involves time series which are terribly valuable for decision-making.  But
if (lacking astute master data management) there is a disconnect in a
definition of an instance in a critical central reference file (e.g. redefining
the boundary of a sales territory starting in April), that discontinuity of
definition can lead to misleading analysis at the executive level. 

This presentation will discuss dataimport, data integration, and managing of change of data for downstream efficacy.  Mr. Scofield draws upon his own lectures about data import andsemantic data integration which have been presented to numerous audiences over the past 14 years.  

Michael Scofield is an Assistant Professor of Health Information Management at Loma Linda University.  He is a popular speaker, writer, and consultant in the fields of data management, data quality assessment, and data visualization.

For more information go to the DAMA Phoenix Chapter home page.