* very data-centric organization * involved in DataGrid
* SOA at AWI [diagram of 2004]
In practice... post-print -> repository PI is supposed to release data when published, but by then, it is lost or there are excuses
So need Staging -> Publication
So: Fedora to do all this.
* Reasons for Fedora - Virtual Repository - not restricted to Dublin Core - standards compliant - etc.
[diagram of 2005]
currently using dc.source and dc.relation as a hack to express linkages, convinced that proper way to do this is RDF/XML
Long-term issues * Benchmarking for large number of files * out-of-box web client acceptance * fine-grained access control and Shibboleth based AuthN - relevant in DataGRID * support for sets * federation model * collaboration and support - disseminators for visualizations services; relevant for DataGrid - Eclipse project to facilitate plugin devel - Google strategy? - seminars, tutorials for advanced Fedora users
Mission - foundation of new and expansion of existing scientific electronic journals - customizable workflow within a common publication system [including peer review] - fast, open and transparent digital peer publishing
Components and Middleware - Peer Review System (external component) - from Peer Review System via LDAP auth to Plone-based Publication System - from Peer Review System via OAI to DiPP Services Fedora-based Repository - from DiPP to Publishing system via SOAP
currently using Fedora 1.2
challenge: external peer review, Plone-based publishing want: at least appearance of one unified system
Publication Pipeline: includes Repository Engine, Conversion Engine, Publication Engine therefore DiPP Services * Conversion Service - uses commercial tool to get to DocBook XML, then XSLT to transform to HTML and even PDF (convert XML to LaTeX using XSLT then to PDF?) * URN service - persistent identifier for journals, articles, supplementary material * Distribution Service - registration for OAI harvesting - RSS feeds - email alerts
also had to build Hierarchy Service because of limitations in Fedora version 1
Fedora - is it good enough?
pros - highly modularized - versioning of datastreams - addition of own metadata formats
cons - added metadata in own formats not searchable - inconsistent versioning of API-A (Access) and API-M (Management) from Fedora 1 to Fedora 2
would be good to have a pool of common services
Summary and Outlook - workflow-based publication system - extended services on top of Fedora
Next Steps - DINI certification (German centre for network information) - full-text indexing via FAST search engine - improvement of usability
Fedora 2.1 - introduce Fedora Service Framework (services distributed as part of the official distribution) guarantee compatibility as Fedora evolves - intro PROAI (OAI Provider Service) ? on sourceforge now ? - intro directory ingest service