I wanted to get myself up to date on contemporary ideas around use of taxonomies vs. folksonomies and was drawn to a course being run by the UKeiG (part of CILIP). The course was led by a renowned and respected information management professional and Fellow of CILIP.
It was like stepping back in time 10 or 15 years, where metadata standards, structured lists, taxonomies, thesauri and controlled vocabularies were paramount in the discipline of effective information management. Discussion on folksonomies, and social bookmarking (the original reason for my attendance) was sadly limited to a brief 10 minute slot at the end of the day. This led me to wonder whether professional bodies such as CILIP had truly grasped the magnitude of the change now taking place in the social computing space, and indeed, whether the social element of information management was recognised at all.
I was reminded of the unnecessarily over-complex government metadata schema e-GMS (a superset of Dublin Core) and the even more complex government subject tag encoding scheme, the Integrated Public Sector Vocabulary (IPSV), now over 8000 terms I’m reliably informed. Though I appeared to be the only one present who understood the connection between ‘over complex’ and ‘poorly implemented’. I thought it was common knowledge that many (most?) departments and organisations in the public sector arbitrarily picked a convenient high-level term from IPSV to classify all their web pages just so that they could tick the box for being IPSV compliant. I wonder how long it’s going to be before this ludicrous standard is consigned to the ‘good idea at the time but impractical to implement‘ bin by the folk over at the Cabinet Office.
Furthermore, I was not convinced by the argument put forward by the course leader as to the benefits of accurate and consistent use of IPSV terms for ensuring good search results. Searching on the term ‘wellington’, could, we were told, return results about the Wellington boot, Wellington New Zealand or the Duke of Wellington.
Right. But if users are foolish enough to use one search term without giving any context, then they deserve to get mixed and irrelevant results. One of the good things about Google is that it has conditioned most people on how to construct reasonably good search queries. I wonder how many users in the public sector would think to themselves as they surveyed their mixed bag of results from the ‘Wellington’ query “mmm, I must contact the webmaster about ensuring IPSV terms are more accurately applied” and how many would refine their search to something like ‘Wellington boot’, Wellington NZ’ or ‘Duke of Wellington’? Indeed, as far as a Google web search is concerned, complete absence of the IPSV meta-tag will make not a jot of difference to the search results because Google know they can’t rely on subject metadata in their search algorithms.
Then at last we finally got to discuss taxonomies and folksonomies. It was clear that folksonomies were not favoured by the course leader, who quickly demonstrated a tag cloud on Flickr, but without explaining why some tags were in a larger font than others (indicating their frequency of use) or the social networking aspect of how the tags got created in the first place. The sole reason put forward as to why folksonomies were not as good as taxonomies for information retrieval was the cost of tagging (?), conveniently forgetting – it seems – that taxonomies also require use of tags.
So, for the benefit of Librarians and other information professionals, and particularly the ones on the course I attended, here is my slightly more detailed analysis of the relative merits of taxonomies and folksonomies:
| TAXONOMIES | FOLKSONOMIES |
| Central control | Democratic creation |
| Top-down | Bottom-up |
| Meaning to the author | Meaning to the reader |
| Process to add new | Just do it |
| Accurate | Good enough |
| Navigation | Discovery |
| Restrictive | Expansive |
| Defined vocabulary | Personal vocabulary |
As always, I am open to other views and opinions from my peers and experts in this field, and in particular I want to be reassured that information management professionals do understand that there is a quiet revolution happening in the world of social computing that threatens some long established standards and practices for effective management of information, and that there are now some new tools in the toolbox.
Just wanted to say HI. I found your blog a few days ago on Technorati and have been reading it over the past few days.
Well if you don’t allow your self to be put off by the off Tweeter and do come on the course, you will find most of day three devoted to lightly constrained tagging systems as an alternative to either of the above ….
It sounds as though the course leader hadn’t really investigated folksonomies much at all or, indeed, spent much time using systems with folksonomies.
My team find that, even when using deep-level IPSV terms, we have to use standard metadata as well. The main reason being that the vocabulary used is very much based on “council speak” and, even if searchable via google, are not words that non-council people would be familiar with.
Not only that but there are changes in terms that are used depending on location both in terms of dialect (eg yorkshire vs cornwall vs essex, for instance) but also legal terms (Scots law vs english law).
All the various standard vocabularies are much more to do with the national government’s project to have a citizen account and being able to find something on one council site in the same way as we would on another.
Also, folksonomies, especially mature ones, are extremely friendly to search engines, because the people creating the tags are the same people that carry out the searches.
ooops, my bad. didn’t realise I was required to put tags.
Dave,
ahh, I should have guessed you would pick up on the Tweeter. I’m certainly intrigued by the course, but as an independent freelancer I have to look careful at cost/benefit. since it’s me that’s paying and not some large corp. Will have a think.
Andrew,
thanks for comments. I’m intrigued to know whether your team use IPSV because it’s mandated, or whether it is adding real value – e.g. for information sharing between councils, or to support document or records management? As you can see from my original post, I can’t see any practical benefit – unless it is tied to something like EDRMS.
Steve,
Mostly because it’s mandated and may be useful in future for connecting simmilar pages across councils. We tend to use the “real” terms in the standard XHTML metadata which google is better at using anyway.
The same applies to all the mandated lists (Navigation list, A2Z list etc …)
We also tend to find that the lists have been made for England. Quite often, Scottish and English law have different terms for the same thing and can even have entirely different set ups.
Even though we have a specifically Scottish Navigation list, the terms used can vary between, say, Dundee and Glasgow. As in England, dialects across the whole country can be very different.
I think, in our case, folksonomy would work really well because council staff can tag the pages with the “council terms” and they can also be tagged with the everyday terms by anyone else.
I have been advocating the Hybrid approach to folksonomies and taxonomies in the Enterprise for a while and recently published an ebook that provides a high level look at the benefits of the Hybrid approach that i hope you find useful and entertaining!
I have been in many a meeting where the mention of ‘folksonomies’ causes visible anguish – but i am seeing that attitude change quickly.
You can download it here:
http://solutions.dowjones.com/cookbook/ebook_sla2008/cookbookebook.pdf
Daniela,
many thanks for the ebook link. A hybrid approach sounds like it could be the best of both worlds. I look forward to reading about it.
Steve