1. A Multi-Discipline Metadata Registry for Science Interoperability J. Steven Hughes/JPL - steve.hughes@jpl.nasa.gov Daniel J. Crichton/JPL - daniel.crichton@jpl.nasa.gov Jason J. Hyon/JPL - jason.hyon@jpl.nasa.gov Sean C. Kelly/UTA - sean.kelly@jpl.nasa.gov Open Forum on Metadata Registries January 17-21, 2000 Santa Fe, New Mexico

2. A Multi-Discipline Metadata Registry for Science Interoperability Background Problem Statement System Overview Profile Development Conclusion and Issues

3. Background NASA’s Office of Space Science Planetary Science Planetary Data System (PDS) 5 Science disciplines nodes - 2 Support nodes 1 Central node Heterogeneous domains - short term missions Astrophysics Astrophysics Data System 100s to 1000s of nodes Homogeneous domains - long term missions Space Physics Space Physics Data System* Several identified nodes

4. Background Planetary Data System (PDS) Archives essentially all science data from solar system exploration missions Prototype - 1986, Operational - 1990 Publishes archive quality products Well defined standards architecture

5. Background Planetary Science Standards Architecture

6. Background Planetary Science Data Dictionary 1000+ Data Elements spanning Planetary Science disciplines Nomenclature Standard Meaning, type, ranges, enumerated values Planetary Science Data Model Developed as Planetary Science enterprise E/R model Planetary Science Entities - Spacecraft, Instruments Science Data Entities - Data Products, Projections,... Data Organization Entities - Volumes Management Entities - Nodes, Personnel Implemented as the PDS Data Set Catalog in an RDBMS Distributed in Object Description Language

7. Background Challenge Develop single interface for locating space science data. Provide data system interoperability. Support correlative Science.

8. Problem Statement Space scientists can not easily locate or use data across the hundreds if not thousands of autonomous, heterogeneous, and distributed data systems currently in the Space Science community. Heterogeneous Systems Data Management - RDBMS, ODBMS, HomeGrownDBMS, BinaryFiles Platforms - UNIX, LINUX, WIN3.x/9x/NT, Mac, VMS, … Interfaces - Web, Windows, Command Line Data Formats - HDF, CDF, NetCDF, PDS, FITS, VICR, ASCII,... Data Volume - KiloBytes to TeraBytes Heterogeneous Disciplines Moving targets and stationary targets Multiple coordinate systems Multiple data object types (images, cubes, time series, spectrum, tables, binary, document) Multiple interpretations of single object types Multiple software solutions to same problem. Incompatible and/or missing metadata

9. Proposed Solution Encapsulate individual data systems. (Hide uniqueness.) Communicate using metadata that describe resources Data (e.g. data sets, images) non-Data (e.g. catalogs, services) Enable interoperability based on metadata compatibility. Refocus problem on metadata development.

10. Proposed Solution (cont) Object_Oriented Data Technology Task (OODT) –Domain independent data management infrastructure Domain independent data structures –XML - Standard interchange language –Metadata management Resource profile –Message passing Domain independent system infrastructure –CORBA for interoperability between computer systems and languages –Message passing to simply interface design –Standardized reusable server components

11. System Overview Object Oriented Data Technology Framework SeaWinds StagingOODT ServerPDS StagingPTI Staging Profile Server Query Server Archive Server Product Server Archive Server Profile Server SybaseOracle Profile Server PDS Systems Product Server Prof Scientist Web Server

12. System Overview Profile Service Profile describes a resource –Available datasets and products –Types of resources and where they’re located Optionally reference other profile servers Profile Server Prof Data system 1 Data system 2 Profile Server Prof Profile Server

13. System Overview Query Service Knows how to “crawl” through servers to produce a result –Crawls through profiles to discover other profiles and product servers –Crawls through product servers to display available products Accessible through CORBA API or through web browser

14. Profile Development Objective Objective Design and develop domain generic structure that will capture the metadata necessary for identifying and locating science data resources across distributed heterogeneous data systems. Result Profile - A resource description (subset of meta-model) sufficient to determine if the resource might resolve a query.

15. Profile Development Approach Choose a common interchange format. Develop a domain generic language. Implement domain specific instances. Model the domain. Capture the meta-data. Develop system to manage the results.

16. Profile Development Choose a common interchange format XML eXtensible Markup Language More expressive than HTML More simple than SGML A meta-language used to define domain languages. XSIL - eXtensible Scientific Interchange Language. XIL - Instrument control language. Wide acceptance as an interchange format. Electronic data interchange (EDI) standard.

17. Profile Development Develop a domain generic language Define a generic structure (XML DTD) that can describe heterogeneous domain-specific resources. Profile - A resource description with sufficient information to determine if the resource satisfies a query. Profile elements name, syntax, unit, value_instance, meaning, alias, … encodes selected domain attributes and their values specific to this resource Resource attributes - id, title, discipline, location_id, … Profile attributes - id, title, desc, type, data_dictionary_id, …

18. Profile Development Develop a domain generic language prof.dtd <!ELEMENT PROFILES (PROFILE+)> <!ELEMENT PROFILE (PROFILE_ATTRIBUTES, RESOURCE)> <!ATTLIST PROFILE PROFILE_ID CDATA #REQUIRED > <!ELEMENT PROFILE_ATTRIBUTES (ID, TITLE*, DESC*, TYPE*, STATUS_ID*, SECURITY_TYPE*, PARENT_ID*, CHILD_ID*, REVISION_NOTE*, DATA_DICTIONARY_ID*)> <!ELEMENT RESOURCE (RESOURCE_ATTRIBUTES, PROFILE_ELEMENT*)> <!ELEMENT RESOURCE_ATTRIBUTES (RESOURCE_ID, RESOURCE_TITLE, RESOURCE_DISCIPLINE, RESOURCE_AGGREGATION, RESOURCE_CLASS, RESOURCE_LOCATION_ID, RESULT_MIME_TYPE)> <!ELEMENT PROFILE_ELEMENT (ELEMENT_NAME, ELEMENT_MEANING*, ELEMENT_ALIAS*, VALUE_SYNTAX*, VALUE_UNIT*, (VALUE_INSTANCE | (MINIMUM_VALUE, MAXIMUM_VALUE))*)>

19. Profile Development Profile Example - PDS Distributed Inventory System PROFILE_PDS_DIS_V1.3.n Planetary Data System - Distributed Inventory System - Profile V1.0 This profile describes the Planetary Data System (PDS) Distributed Inventory System (DIS)... PROFILE OODT_PDS_DATA_SET_DD_V1.0 PDS_DIS_V1.3.n Planetary Data System - Distributed Inventory System PDS GRANULE+ INVENTORY http://pds.jpl.nasa.gov/pdsbrows.htm text/html...

20. Profile Development Profile Example (cont) - PDS Distributed Inventory System … DATA_OBJECT_TYPE The data_object_type element provides the type... ENUMERATION N/A IMAGE... DATA_SET_NAME The data_set_name element identifies a PDS data set. -- example... ENUMERATION N/A VO1/VO2 MARS VISUAL IMAGING SUBSYSTEM DIGITAL... VO2 MARS RADIO SCIENCE SUBSYSTEM RESAMPLED LOS …... TARGET_NAME The target_name element provides the names of the targets... ADS.OBJECT_ID ENUMERATION N/A IDA JUPITER...

21. Profile Development Develop a domain generic language Specialize the profile class Profile - One profile to one resource (e.g. inventory) Inventory - One profile to many resources (e.g. data set, image) Minimized profile element attributes no meanings subsets of preferred values Dictionary - One profile to one discipline Maximize profile element attributes aliases, meanings union of all preferred values

22. Profile Development Develop a domain generic language Profile element hierarchy Dictionary - Planetary Science Data Dictionary data elements - union of all data elements in all profiles preferred values - union of all data element values e.g. TARGET_NAME = {ADRASTEA, …, VENUS} Profile - Planetary Image Atlas - Viking, Galileo, MPF,... data elements - union of all data elements for all entities managed by resource preferred values - union of data element values e.g. TARGET_NAME = {MARS, DEIMOS, PHOBOS, JUPITER,...} Inventory - Viking Orbiter Image Catalog data elements - data elements associated with inventory item. perferred values - data element values for inventory item. e.g. TARGET_NAME = {MARS, DEIMOS, PHOBOS}

23. Profile Development Implement domain specific instances Apply domain generic language to specific domain. E.g. Space/Earth Science data and other resources. Model the domain Data Dictionary Data Model Capture the meta-data Extracted from domain metadata repository

24. Profile Development Implement domain specific instances Inventory Example - PDS Data Set VO1/VO2-M-VIS-5-DIM-V1.0 VO1/VO2 MARS VISUAL IMAGING SUBSYSTEM DIGITAL IMAGING MODEL... PDS GRANULE+ DATA http://pds.jpl.nasa.gov/cgi-bin/pdsserv.pl?OBJECT_ID=PDS100676... text/html DATA_SET_NAME VO1/VO2 MARS VISUAL IMAGING SUBSYSTEM DIGITAL IMAGING MODEL... DATA_OBJECT_TYPE IMAGE TARGET_NAME MARS VOLUME_ID VO_2001... VO_2014

25. Conclusion Profile Development - Review Choose a common interchange format. (XML) Develop a domain generic language. (X2PL) (XML eXtensible Profile Language) Implement domain specific instances. (Resource Profiles) Develop system to manage the profiles. (Profile Servers)

26. Conclusion Issues Develop space science metadata registry ~10 high level concepts - “Anchor Points” Complete development of discipline registries Determine management policy Design meta-model and mandate conformance Evolved meta-model through voluntary conformance Determine space science metadata standards NASA Data Entity Dictionary Specification Language (DEDSL - XML syntax) currently being used