Suggestions for improving refbase:

- Open up the refbase group (aka tagging) mechanism so that groups can be made
  public, group-specific or private.

- Improved search capabilities which, e.g., allow for 'OR' searches and/or a
  Google-like search syntax. One possible solution might be to offer support for
  the Contextual Query Language (CQL) across the entire interface. More flexible
  handling of accented characters and umlauts would be also desirable.

- Offer means to auto-detect duplicates on import, link duplicate entries with
  their original ones and allow users to migrate their user-specific data from a
  duplicate record to the original. It would be also nice if duplicate records
  could be hidden from the interface.

- Improve the installation process so that the web-based installer works with
  more server setups (including installation on shared hosts), and allows
  setting important config options via the web interface.

- More user-specific and record-specific rights management.

- Provide additional citation styles, preferably via the use of a plugin
  mechanism which would make it easy to add or edit citation styles. Support for
  the Citation Style Language (CSL) would be desirable.

- Increase number of supported resource types and allow users to add their own
  types. A hierarchical system with basic fallback types would be desirable.

- Provide a machine-readable interface (API) that not only allows clients to
  query a refbase database and retrieve results in a structured format, but that
  also allows modification of any database information.

- Provide better integration with desktop reference managers so that users could
  search an online refbase database from within their desktop application, or
  upload (and eventually synch) records to a refbase server directly from within
  their desktop application.

- Upon add/edit, offer to auto-fill empty fields based on information available
  in other fields/records/tables or via web services. Eventually, such a system
  could also be used to validate data that were entered/imported by the user.

- Simplify manual data input by providing record type specific forms. These
  forms could guide a user by only showing (or at least indicating) those fields
  that are required for a particular record type (like journal articles, book
  chapters or whole books).

- Allow users to upload multiple files per record.

- Allow users to specify multiple URLs for the same record, and eventually
  provide a customizable system for auto-generation of links (based on the
  record's metadata).

- Store record identifiers from foreign repositories (e.g. arXiv.org or
  PubMed.gov) in some dedicated field(s), which would allow to easily
  auto-generate links back to these repositories.

- Implement the OAI Protocol for Metadata Harvesting which would provide (yet
  another) standard way for software to retrieve information from refbase.

- More localizations, i.e. translations of the refbase interface into other
  languages.

- Cross-site searches (both through other copies of refbase and through other
  bibliographic databases).

- Refactor for more code reuse (share more between cite/list/details views,
  etc.), and eventually modularize the code to allow for other RDBMS, custom
  database schemes and/or a plugin mechanism.

- Provide better interface customization via a templating/skin mechanism and/or
  improved CSS styling.

- Allow the admin to add any custom fields and/or alter the layout of available
  fields.

- More embedding of citation metadata in HTML pages in a machine-readable form
  (such as embedded RDF or microformats) which may be processed by clients.

- Improved export through both revisions to the current MODS XML/bibutils system
  and through adding more native export options which wouldn't require bibutils.

- User customization: provide a user-specific preferences page where one could
  specify the default view and the default columns that are visible in query
  results or the default fields that are visible in particular search forms.

- Easier import of bibliographic data from web sources via bookmarklets and/or
  scrapers.

- Offer a notes/comment system where multiple notes (of unlimited length) would
  be allowed for a particular record and where notes would be searchable, and
  could be made public, group-specific or private.

- Enable remembering of marked records across different requests.

- Provide an administrative interface to deleted records.

A more detailed list of planned feature additions is given at:
  <http://additions.refbase.net>