Visiting the Invenio Team at Cern

Invenio and authority files

Controlled lists like the one for journals used in infoscience could be modelled using the invenio knowledge base concept. This would provide for normalization when outputing data. Since the data is not normalized in place, this approach doesn't help for searching. 

Invenio and access control

As of now invenio supports locally defined groups. But the implementation of support for external groups - e.g. defined in an ldap server - is under way. Once this is implemented, relation of users with labs could be stored as user specific settings on login, and probably used for access control. 

Invenio and harvesting/receiving data from labs

The preferred way of synchronizing infoscience and separate publication databases would be via oai pmh. Invenio provides a module implementing the harvester side of the protocol to integrate external data. As an alternative to the dropbox, the email-submission functionality of invenio could be used with labs. 

Curator vs. Development at Cern

Two modules of cds invenio (one not included in public cvs yet) overlap with the envisioned functionality of curator: 

• BibCheck - correctness/completeness check of data in MarcXML 

• http://cdsware.cern.ch:8000/admin/bibmatch/guide.html

  • duplicate checking. 

But even when using those modules it would still make sense to develop a web interface for them - since they will provide command line tools. 

Summary

Since the invenio team is interested in codevelopment, too, upgrading infoscience to work with invenio from cvs should be the highest priority.  To do so, one would have to start with the software in the infoscience development branch, because the production branch is too far away from current invenio. Since the development branch is not fully functional yet - see bugzilla - this also means some development/backporting of functionality which is already implemented in the production branch. If the migration of the data succeeds and can be automated, this procedure should be ok. To make codevelopment with the cern easier, infoscience and invenio should be as loosely coupled as possible (preferably no imports of invenio functionality from infoscience). 

RobertForkel/2007-02-07 (last edited 2011-06-03 06:35:27 by localhost)