Metadata Frameworks

From NGDCWiki

Jump to: navigation, search

This page will serve as a scratchpad for evaluation of Fedora, ebRIM and GeoNetwork. It's a long ways from a finished product for public consumption!

Three new technologies, Fedora, ebRIM and GeoNetwork, have been proposed as applicable to building a new ISO 19115+ metadata repository for scientific metadata in NOAA/NESDIS.

These may be seen as competitive or symbiotic technologies. They have at their base an XML representation, and are in use for traditional library management and for scientific data sets. They are open source Java/Tomcat, and based on open standards. My first impression so far is that Fedora has a more mature and robust implementation, and more complete create/edit capabilities, while ebRIM offers some additional standard discovery interfaces. GeoNetwork is a more specialized application to ISO metadata, while Fedora and ebRIM are more general-purpose object management tools. It may be possible to combine them, e.g. Fedora or GeoNetwork is the base system, while ebRIM may be another "view" into the same metadata.

Contents

GeoNetwork

Standards based Geographic Data and Information Management System for the web

Background

Catalog application to manage spatially referenced resources through the web. Metadata editing and search functions as well as an embedded interactive web map viewer.

GeoNetwork had a strong presence, including a nicely packaged DVD for the FOSS conference in Victoria, CA, Sept 2007. It seems to have active developer and user communities, including the UN FAO (Food and Argiculture), and robust sponsorship from Autodesk et al. Last release (as of 11/07) was 2.1 (9/07)

Concepts

  1. Search access to local and distributed geospatial catalogues
  2. Archive supports data, graphics, documents, pdf,...
  3. An interactive Web Map Viewer to combine Web Map Services from distributed servers
  4. accessible as RSS
  5. Online editing of metadata
  6. ISO19115/ISO19139, FGDC and Dublin Core
  7. Scheduled harvesting and synchronization
  8. access control, Group and user management
  9. Multi-lingual UI

...

Technology

  • A Tomcat or Jetty (built-in) webapp
  • Can use a built-in (McKoi) database, Oracle, Postgres or MySQL.
  • built-in Z39.50 server
  • CSW 2 server (w/ optional client)
  • Geoserver web map server for WMS, WFS, WCS, OpenLayers, ArcIMS
  • Java 1.5, XSL, web services, JDBC
  • Tested/maintained on Windows, Mac OS X, Linux
  • Jeeves Java/XSL/JDBC engine


...

Links

Fedora

Flexible Extensible Digital Object and Repository Architecture

Background

Fedora is a digital object management framework started in 1997 at Cornell, and was joined by UVa in 1999 for their open-source Digital Library project. There is a sourceforge site, frequent and recent releases, dozens of academic/scientific users (including the local CARL libraries), and a recent $5M grant with new software and outreach (marketing) hires.

Concepts

  • XML serialization of objects
  • content versioning
  • Disseminators (methods) for different content views
  • REST-based introspection of objects
  • Dublin Core metadata for all objects
  • Datastreams (Items) can be internal XML or external URls
  • bDefs and bMechs
  • object relationships use RDF syntax (subject-predicate-object)

Technology

  • A Tomcat webapp
  • Can use a built-in (McKoi) database, Oracle, Postgres or MySQL.
  • The internal datastore is XML, so a separate application (e.g. ebXMLrr) can access the same digital objects.
  • Can import XML to define objects
    • presumably we could use XSLT on the ISO 19115 schema to generate them
  • REST and SOAP interfaces are provided
  • Web GUI (Fez) is available.
  • RDF is used to specify relationships between objects.
  • OAI-PMH used to serve object metadata

Links

ebRIM

ebXML Registry Information Model (e-business XML)

Background

ebXML is a collection of web service standards for the e-Business world.

  • ebXML Registry (ebReg) Is to B2B apps what databases were to enterprise apps
    • ebXML Registry Information Model (RIM) or simply ebRim
    • ebXML Registry Services (RS) or simply ebRS

Concepts

In ebRIM, proxy Registry Objects (RO) are created for each Repository Item (RI) in order to support indexing/discovery. The RI's may in

  • the Registry itself
  • in the Registry Repository; for static docs (ExtrinsicObject)
  • external to it (ExternalLink)

Associations and Classifications are also RO's. Association types are Classification nodes. Association source and target are RO's.

RO's can be created by ingesting from another metadata XML; we will need a mapping from ISO (and FGDC?). See mapping examples at https://www.seegrid.csiro.au/twiki/bin/view/Compsrvices/RegistryService

RO's are subsets of the ingested metadata, so it's not clear how we get ISO md out, nor how ISO metadata can be created/edited (Lifecycle Manager?). Still looking into this...

v3 of ebRIM which has the following features:

  • coupled Registry and Repository
  • discovery interface (Query Manager) including support for stored queries
  • registry maintenance interface (Lifecycle Manager)
  • subscription and event notification model
  • support for certain kinds of content validation
  • registry federation mechanism (cooperation among multiple registries to provide unified service)
  • security and authentication model

There is mention of an ebRIM profile of CS-W -- not sure how mature this is.


Technology

freebXML ebXMLrr is a open-source free implementation of the ebREG specification. It includes both server and client implementations that are adheres to the v3.0 specification.

  • Server: tomcat-based and uses Sun's JAXR (for XML Registries) and JWSDP (Web services)
  • Clients: java-based application (Registrybrowser) and a web-based "thin" client
  • Database: uses embedded Derby or HSQLDB database; other databases require clean build (?). Tested w/ Oracle and Postgres
  • Projects using freebXML: NIST, NIH, Sun ,...

Sun's Service Registry Integrates ebRIM and UDDI for WSDL management. Open source, part of Sun Java Enterprise System (based on ebXMLrr?)


Links

information model in UML format to the ebXML Registry Information Model. January 19, 2006.

Personal tools