There’s a lot of talk in the blogosphere right now about folksonomies (user defined keyword tagging). Some of the key issues that have cropped up are that there is absolutely no quality control leading to spelling errors and duplication, tags which are only meaningful to the individual, and internationalisation issues. Tags can be nested, but there’s no thesaurus structure (broader term, narrower term, used for, related term, etc), and not faceted as librarians know it. Yet despite these problems, folksonomies allow personalisation, flexibility, and a greater level of specificity than was possible with existing thesauri.

There’s discussion about fixing up some of the most obvious problems, like spelling and how this would be solved either by sites with tagging or by individuals. To me, the issue of folksonomies and tagging remind me of HTML issues of around 1997. Many experts were up in arms at the number of websites using blink tags and other horrible hacks to render their pages in non-standard compliant code. Eventually the consensus was, you can’t teach everyone to do it right, if you allow everyone to create without restriction. It’s the same with tags – you can’t teach everyone to create better tags. This is one of the reasons why for many years, libraries relied on authority-created, pre-coordinate indexing (choosing terms from a set list) rather than post co-ordinate indexing (similar to folksonomies – choosing terms from the context, though chosen by experts not end-users). But when you restrict the ability to create tags, you gain authority but lose the ability to incorporate new terms quickly – for example, “folksonomies” itself didn’t exist a couple of months ago.

Moving on from the way folksonomies are defined, another major issue is that of internationalisation. Some are looking for an way to map tags from one language to another, despite the mammoth scope and complexity of such a task. People use words in different ways in different languages. Languages are constructed differently from each other. It’s the reason why auto-translators like Babelfish don’t work very well, because when you factor in tenses, politeness, context, slang, and so on, translation is a long and difficult process. So it’s unlikely that there ever would be a perfect map across all tags.

What are the implications for libraries? To me, folksonomies combine a little of what librarians already do, and add another layer of access. Those of us dealing with collections that aren’t well-described by LCSH or DDC turn to thesauri and taxonomies to fill the gap and provide meaningful access to users. At my library, we provide access to music. We found taxonomies for genres and instruments to give users more information. But there are still gaps – when a user asks for “slow, dreamy beach songs” we don’t have any terms from the taxonomy to find that. But, if we used folksonomies, we could. Jenny Levine sees potential in folksonomies being used to create more user-centred services. As she says –

“So how could we make better use of the integration of folksonomies and user-based vocabularies? I’m not suggesting we throw the bath water out with the baby, because I’m also a big fan of structured searching, and let’s face it - one of the things Google isn’t good at is searching structured data. But why can’t we offer both? They aren’t mutually exclusive.”

Absolutely! I think we’re in for a renaissance of cataloguing in 2005.