Comments on: Creating useful classifications with taxonomies (part 1) http://www.michelepasin.org/blog/2013/07/25/creating-useful-classifications-with-taxonomies-part-1/ At the core of all well-founded belief lies belief that is unfounded - Wittgenstein Wed, 05 Feb 2014 12:26:37 +0000 hourly 1 https://wordpress.org/?v=5.2.11 By: Mikele http://www.michelepasin.org/blog/2013/07/25/creating-useful-classifications-with-taxonomies-part-1/comment-page-1/#comment-14424 Sun, 11 Aug 2013 22:07:00 +0000 http://www.michelepasin.org/blog/?p=2389#comment-14424 Hey Geoffroy, thanks for your comment. I totally agree, creating a good taxonomy is a joint endeavour between domain experts and classification experts. Defining a methodology upfront helps – however it’s very easy to run into boundary cases that require adjusting it.

As for the official definition of taxonomy, I think I did run into something like that in the past; I’ll dig it up and maybe include it in the follow-up to this post :-)

]]> By: Geoffroy http://www.michelepasin.org/blog/2013/07/25/creating-useful-classifications-with-taxonomies-part-1/comment-page-1/#comment-13695 Mon, 29 Jul 2013 10:12:00 +0000 http://www.michelepasin.org/blog/?p=2389#comment-13695 I also think it’s good to remind this rule. I’ve encountered this issue in a project where data editors could add new concepts to the taxonomy use to categorise their content on a website. The first two initial levels where consistent but concepts added later at deeper levels tended to rely on increasingly loose variations of the meaning of the original relationship.

As a rule of thumb I recommended that each time they add a new concept C under B they ask themselves the question is C a type of B? and the same question for C with respect to every parent of B in the tree.

This was a good example where freedom was given to domain experts to control their taxonomy and let it grow more quickly but the integrity of the taxonomy must be monitored by someone who knows better about the general principles of ontologies. However I believe that, in practice, the main difficulty comes from the fact that both types of knowledge (about the content being categorised and the rules governing the conceptual structure holding this categorisation) are not completely separated.

Is there an accepted formal definition of a taxonomy in information science? Could we say, for instance, that it is a formal ontology using a single type of relationship (which has to be transitive and antisymmetric) and takes the form of a tree.

]]>