110 likes | 239 Views
This resource explores effective modeling paradigms for the Data Documentation Initiative (DDI) as presented by Dan Gillman from the US Bureau of Labor Statistics. It emphasizes various modeling methodologies such as Entity-Relationship (ER), Unified Modeling Language (UML), Object-Role Modeling (ORM), and highlights the importance of the Generic Statistical Information Model (GSIM) and the Generic Statistical Business Process Model (GSBPM). The text discusses the life-cycle support, data set management, and the need for clear specifications in survey methodology and cognitive psychology.
E N D
Tips for Building an Effective Model for DDI Dan Gillman US Bureau of Labor Statistics
Modeling Paradigms • Entity-Relationship (ER) • Older • Used to design RDBs • Looser rules • IDEF1x • Based on ER and RDBs • Tight rules • ISO/IEC 31320-2 • Not bound to software development
Modeling Paradigms • Unified Modeling Language (UML) • Based on OOD and OOA • ISO/IEC 19501 • Based on MOF • Tight Rules • Bound to Programming Languages (Java, C++, etc) • Major support by other specs • Interchange models through XMI
Modeling Paradigms • Object-Role Modeling (ORM) • Based on NIAM (built in 1980s) • Purely Conceptual Modeling • Fact-Based Modeling • Diagrams to be read as series of statements • Similar to RDF • Graph like structure • Separates Concepts and Objects • Not so much for system design • Best for understanding
Current Situation • XML-Schema • Modeling language as well • Designed to be human readable • Not so easy • Very verbose • Relationships • not naturally supported • Overall design not easily illustrated • Limits implementation types
Current Situation • Current DDI • Much good work • Good statistical understanding • Life-cycle support • Data set support • Shouldn’t be lost • But not the only game in town
GSIM • Generic Statistical Information Model • Version 1.0 due in Dec 2012 • Version 0.8 out for review until 19 October • Companion Model to • Generic Statistical Business Process Model • GSBPM • GSIM – data model • GSBPM – process model
GSIM • Detailed model for • Variables, Data, Classifications • Data sets • Processes and Activities • Full support for life-cycle
DDI/ SDI • Survey Design and Implementation WG • Questionnaire design • Frame and Sample description • Weighting • Paradata
Design Criteria • Terminology (Concepts and related things) • Methodology • Statistics and Probability • Cognitive psychology • Computer science • Computation (IT resources) • Normative specifications • Rationales • Specifications • Procedures
Level of Detail • Multiple views • Conceptual layer • Managers • High level view • Communication layer • More detail • Still conceptual • Specification layer • Implementation and interoperability • Detailed metadata