With or Without UIDs

In a world where information and media flow freely, being able to share reading lists, playlists, recommendations and citations is important. Many of these items have little metadata attached, but this metadata can still be shared in standard, re-usable formats: XSPF, microformats, RDF[a], OpenURL/COinS, etc. The latent information present within the objects, such as audio fingerprints, can also be collected. Then we can use services – scrapers, analyzers, resolvers, search indexes, disambiguation and normalisers – to connect them to identifiers: DOIs, MusicBrainz IDs, ISBNs, PubMed IDs and canonical URIs.

There are many similarities between the ways that different kinds of digital objects, such as music or academic papers, are handled. For example, a researcher might gather a collection of academic papers on a particular topic. As they add items to the collection, their reference manager (Zotero, say) will analyse the documents and web pages, using scrapers and web services to fetch and convert as much metadata as possible, and attach identifiers. If they then share that reading/reference list with someone else, the recipient should be able to use those identifiers – or even just the basic metadata, using OpenURL, xISBN and other resolver services – to fetch a copy of the object from wherever is most appropriate.

Similarly with media playlists, all the available metadata can be used to identify files as they’re added to the collection. These identifiers can again be used to pull in more metadata, which is all useful for automatically building playlists, making recommendations and displaying contextual information (lyrics, album covers, events, photographs, etc). When these collections are shared as XSPF playlists, the recipient can use these identifiers and metadata to resolve and fetch an appropriate copy of the files to which they have access. This process also works in reverse – when a media playlist lacking identifiers is shared, the recipient can use appropriate resolvers to look up identifiers and from there connect the individual digital objects to the full network of information.

Finally, a discussion of how web-based tools and desktop applications can communicate. By running embedded HTTP or XMPP servers, desktop applications can be notified of new events occurring online; similarly, by the use of extensions, web browsers can be enabled to communicate with these desktop applications from within their shifting interfaces.