Taxonomies and Classification for Organizing Content Prentiss Riddle

Taxonomies and Classification for Organizing Content Prentiss Riddle INF 385 E 9/21/2006

What is a taxonomy? • animal A hierarchical – vertebrates classification system in • mammals which categories are – cat subdivided to create – dog • reptiles finer distinctions. – invertebrates • vegetable • mineral

Not all classification systems are taxonomies Richard Saul Wurman’s LATCH • • • Location Alphabet Time Category Hierarchy Any of these can be expressed as a taxonomy or not. (Well, maybe not the alphabet. . . )

A quibble about definitions • We commonly use “taxonomy, ” “hierarchy” and “classification system” interchangeably but in fact they’re distinct (if overlapping) terms • In particular, a hierarchy can be linear – the “H” in Richard Saul Wurman’s LATCH – elephant > horse > dog > mouse – Maslow’s hierarchy of needs • A “folksonomy” is not a taxonomy!

Why taxonomies? Taxonomies in our heads • Fundamental to cognition • Observed in children from an early age • Long before Linnaeus and Darwin: “folk taxonomies”

Why taxonomies? Taxonomies in the world • Genealogy, phylogeny • Command structures, org charts, territories • Filesystems, domain names, URLs /var/www/people/faculty. php www. ischool. utexas. edu http: //www. ischool. utexas. edu/people/faculty. php

This suggests the use of taxonomies in IA • Taxonomies in support of browsing • Taxonomies in support of search But that’s problematic. . .

Problematic in the world • The world is complicated Who here has had cataloguing? LC Subject Headings, LC Classification, Dewey. . . • The world is even more complicated – Lattices and networks, not trees – Multiple kinds of relationships – Fuzzy boundaries

Problematic in our heads • The taxonomies in our heads don’t match the world • The available taxonomies may not be what’s important anyway – The user doesn’t care about the org chart! (The mantra of enterprise IA) – The user doesn’t care about the filesystem

Cautious use of taxonomies • Tempered by understanding and testing users • Enlist users in creating taxonomies – Techniques like card sorting • Functional or “folk” etymologies preferred over official ones – A gardening site might classify plants by temperature, sun and water needs, not by botanical classification • Make official taxonomies available behind the scenes for use by experts (departmental admins, biology wonks)

Taxonomies in support of browsing “Umbrella architecture” (Rosenfeld & Morville) E. g. , familiar hierarchical menu structures

Taxonomies in support of browsing Not always ideal. . .

Taxonomies in support of browsing. . . but better than chaos. • Supports breadcrumbs • If the users don’t know your taxonomy, gives them a fighting chance to learn it.

Browsing very large taxonomies Yahoo began as a taxonomy company. . . although at some point it became a search (and content) company

Taxonomies in support of search Search results can include a link to a category

Taxonomies in support of search And searches can be limited to a category

Beyond taxonomies

Taxonomies on steroids Taxonomy + controlled vocabulary = thesaurus A thesaurus usually has a taxonomy embedded in it

Relationships in a thesaurus • Some of the links in a thesaurus express hierarchy and links across it – Broader term (BT) – Narrower term (NT) – Related term (RT) • Some express the controlled vocabulary – – Preferred term (PT) Variant term (VT) Use (U) Used for (UF)

Not just synonyms • A thesaurus is not just for synonym rings cat = feline = kitten= kittycat • It’s also for key relationships across the hierarchy “Nice pants! How about a shirt? ” In a sense, Amazon’s many suggestion features and much of Google Adwords are a set of RT links in a thesaurus

Faceted classification A problem inherent in taxonomies is, what gets divided first? – History - U. S. - War - 1812 – War - History - U. S. - 1812 – U. S. - History - 1812 - War Or in more familiar terms: – Wine - Red - California - Dry - 1999 - Under $25 – Wine - Under $25 - Red - Dry - California - 1999 – etc.

Facets are independent hierarchies • Facets work in parallel – In the subject classification example: facets for topic, time, place, etc. – In the wine example: facets for type, origin, price, manufacturer, etc. • A particular item will be at the intersection of several facets • Facets can simplify classification systems both for creators and users

Folksonomies • Fun, powerful, interesting but a misnomer: not taxonomies at all • Tagging systems in use in popular “web 2. 0” sites – www. flickr. com – del. icio. us • Personal keyword metadata aggregated for searching and browsing • The result is not a hierarchy, not really a classification system, certainly not a taxonomy