This is probably going to make me extremely unfashionable. I know it's not cool to mention 'taxonomy' these days.
Many of us have embraced the idea of folksonomy approaches to content classification (unstructured keyword tagging by information users) as opposed to what we see as structured formalised classification systems (taxonomies created by information experts).
I wouldn't try to deny that there are massive benefits to a folksonomy approach to organising content:
1. It's user centred and more objective (how information users see content)
2. It does the job of building the classification system and connecting it to content at the same time
(rather than a) build the taxonomy b) try to connect it to content)
3. Delegating the job of classification to users gets the job done quickly where information resources are industrial scale
But should we go as far as Emanuele Quintarelli to say that taxonomies are limited in their usefulness only to small specialised repositories of information? http://www.iskoi.org/doc/folksonomies.htm
Only if we assume - that information users have low expectations of consistency and comprehensiveness in their search results. Keyword tagging, only works for a large corpus and we mean vast because it relies on the same user satisfaction principles as generic search engines: users only ever retrieve a fraction of relevant results, but they are happy because a) they don't understand the size of the corpus or the quality of what's missing b) there is so much retrieved anyway that its more than enough to satisfy their hunger.
However search engine trends show us that users are navigating towards a growing number of specialised search engines or search engines that search across search engines www.rollyo.com or www.goshme.com that retrieve more of the relevant quality results than general search engines.
Elaine Peterson brilliantly sums up the philosophical and practical issues with folksonomy tagging
http://www.dlib.org/dlib/november06/peterson/11peterson.html
I would argue that a more formal information architecture: whether taxonomy, ontology, topic map or semantic wordnet does far more to advance user experience that merely to enforce constraint.
In all of the above, the keyword tags are formed into a structure that has meaning. Meaning that a computer can understand and employ to help users navigate specialised AND vast repositories of information. This embeds more artificial intelligence into the search environment and creates a far more sophisticated user experience.
The argument for ontologies is that they are a knowledge resource in themselves and as an information user they help me to navigate information because they understand the relationship between terms and overcome many of the semantic issues described by Peterson.
An ontology:
1. is a knowledge base in itself, explains the meaning and relationships between concepts
2. supports information seekers with different levels of knowledge to understand terminology
3. ensures consistent and comprehensive retrieval
I,m currently exploring the use of folksonomy in an enterpise environment as a source for developing user centred ontology.
Comments