Notes

Outline

Current Directions for GovStat Metadata Efforts Carol A. Hert December 5, 2003

Where we are Extensive knowledge of how and what metadata supports finding and understanding tasks Metadata on time, geography, topicality Relationships among entities Modeling that knowledge as DTD/Schema for integration into SKN Relationship to ontological structures Increasing knowledge of agency metadata and DDI activities

Directions to Pursue Our DTD/Schema’s role in the architecture of the SKN “Business rules” associated with metadata and its use

DTD/Schema in the SKN: Key Issues How does data/metadata flow into and out of the DTD (from agencies, to tools)? Mappings to support Communications protocols Human dimensions: mark-up, cataloging rules, legacy data, etc.

DTD/Schema in the SKN: Key Issues (2) What other SKN components will need metadata from the DTD or will support the DTD? (e.g., ontologies, geospatial mapping tools, relation browser, SIG) What software is necessary to manipulate DTD/Schema information? (e.g., parsers, information retrieval tools, tools for DTD management activities)

Metadata Business Rules These rules are both intra- and inter-DTD How will information within the DTD be manipulated, what content needs to be standardized, and so on? How will information from multiple entities be integrated? Supporting both technical and human dimensions of comparisons

Metadata to Support Comparing Activities Comparisons a critical user activity and SKN functionality issue Some types of comparisons SKN needs to facilitate: Across geographic units Definitional differences across concepts and variables Methodological differences Across different sources (websites, censuses, surveys, reports) Across units of time

Current Activities Identification of comparison scenarios Foreign labor markets Unemployment numbers Response rates Interviews/document analysis to gather expert knowledge on dimensions of comparisons Synthesizing knowledge in the form of Use Cases For guidance in SKN modeling efforts

Some Tantalizing Early Thoughts Assessments of comparability involve comparing methods, concepts, scope, time periods, geographic coverage May be possible to perform one-to-one comparisons across elements in DTD Depending on the task, data may be considered comparable to a greater or lesser extent

Some Tantalizing Early Thoughts (2) Specific “red flags” exist Seasonal adjustment If “comparable” numbers don’t have face validity, extend the picture you are looking at with additional data Knowledge of domain Tools to facilitate comparisons Exploratory data analysis tools

More Information Metadata Research Agenda paper coming out in Spring in Social Science Computing Reviews “Statistical Metadata Needs during Integration Tasks” paper presented at the 2003 Dublin Core Research Meeting Cahert@syr.edu