Connotea (http://www.connotea.org/) is a new, free, online information management service for scientists created by Nature Publishing Group. It aims to take the social bookmarking approach pioneered by del.icio.us (http://del.icio.us/) and apply it to the tasks of reference management and discovery faced by researchers on a daily basis.

The scientific information space offers some unique opportunities for adding new and interesting layers on top of basic social bookmarking. Connotea demonstrates three enhancements: extraction of additional information for bookmarks, marrying of unstructured and structured data, and rich use of identifiers and linking. This presentation describes these and discusses possible future implications and developments.

Firstly, Connotea makes heavy use of URL scanning. Scientific publications often offer structured data about their articles. Wherever possible, Connotea supplements a bookmark with this bibliographic information—data such as author names, publication date and article title. Every URL that is posted to Connotea is analysed to determine whether Connotea knows how to collect additional data from that source. For example, PubMed (http://www.pubmed.org/) is a large repository of biomedical abstracts, and offers a simple web service interface to the bibliographic data associated with each article. A PubMed article URL contains a PubMed ID, which is recognised and extracted by Connotea during the URL scan and used to query the PubMed web service, all in real time.

Secondly, Connotea attempts to marry the worlds of unstructured and structured data. It uses the additional bibliographic data it collects to supplement tagging as an information retrieval mechanism. As well as using tags to organise and navigate their collections, users can perform text searches on the bibliographic information. We found that a search interface into this data made more sense than a link-based approach, as it subtly separates personal from authoritative data, and avoids naming ambiguities. In other words, tags can be clicked on, but article author names or journal titles must be searched for instead.

Thirdly, Connotea makes use of the linking and identification technologies that have been developed for academic web publishing. It uses DOIs (http://www.doi.org/) to link bookmarks to their original sources. For example, a bookmark for a PubMed abstract would link not just to the PubMed entry, but also to the the full text of the article on the relevant publisher’s site. It also uses OpenURL (http://www.niso.org/committees/committee_ax.html) to link to library holdings catalogs.

Finally, Connotea offers the basic social discovery mechanisms of ‘related tags’ and ‘related users’ to enable users to find articles and researchers in similar fields. It also offers user groups in order to facilitate collaboration.

Connotea has been able to use social bookmarking and tagging to greatly enhance web-based scientific information management, but in doing so has highlighted some future challenges. In particular, the developers must write new rules for each individual site we wish to automatically collect reference data from—there is surely scope for a standard way of explicitly and transparently linking web resources to data about those resources. Such a mechanism would have applications far wider than reference management tools.