What is Facet Analysis?

Facet analysis is the process of breaking down a subject into its component parts, choosing appropriate terminology to express those parts, and conveying those aspects by means of a notational device (Spiteri, 1998). This method was originally devised by S. R. Ranganathan (1962) when describing colon classification, a faceted classification scheme. Faceted search is made possible by facet analysis.

To hear some DAM professionals talk about it, faceted search is the best thing since buttered toast. With good reason too. In a faceted classification structure objects can ‘exist’ in several different locations versus being forced into a single node as in a hierarchical structure. Documents become easier to find in a faceted taxonomy because they are indexed exhaustively, allowing users to search results according to a range of qualities. In effect, faceted organizational structures flatten complex taxonomies allowing collections of digital assets to be browsed.

Examples of faceted classification:

Art and Architecture Thesaurus® Online
http://www.getty.edu/research/tools/vocabularies/aat/

Library of Congress Classification System
http://www.loc.gov/catdir/cpso/lcc.html

Examples of faceted search:

Amazon Instant Video (UK)
http://www.amazon.co.uk/Instant-Video/

National Film Board of Canada (films by subject)
https://www.nfb.ca/subjects/

In the field of information science, the output of facet analysis is usually a taxonomy that displays relationships between terms to be used in subject classification of digital assets (see Boxes and Arrows for an example). Single concepts must be identified and distinguished from one another. These concepts may be post-coordinated (synthesized) at the point of information retrieval.

Example of faceted terms:

Human

(gender)

Male
Female

(nationality)

American
French
German
Russian
etc.

Example of post-coordination:

Male – Russian

According to Ranganathan (1967), facets must be mutually exclusive, homogenous, reflect the purpose, subject and scope of the classification system, and be exhaustive. In other words facets should completely describe the subject using terms that are distinguishable from one another. The index terms should also reflect current usage in the subject field.

Generally speaking, there are five steps to facet analysis.

1. collect a representative sample of the assets
2. create a one sentence statement that represents what each item is about
3. extract indexable concepts (entities) from these statements
4. break apart any compound subjects
5. organize into categories

Ranganathan (1967) came up with five fundamental categories (PMEST):

1. Personality: something or someone in question (Who)
2. Matter: aspects personality is made of (What)
3. Energy: how personality changes, is processed, evolves (How)
4. Space: place, location, environment personality exists (Where)
5. Time: dates, periods, etc. action takes place (When)

If you’re left needing a broader set, Broughton (2001) came up with thirteen (more granular) categories.

• thing/entity
• kind        
• part
• property
• material
• process
• operation
• patient
• product
• by-product
• agent
• space
• time

Or make up your own!

There exists no categories that are fundamental to all subjects. Categories should be derived based on the subject being analyzed (Spitery, 1998).

The resulting structure will be the basis for a taxonomy. Next up: organize, test, deploy, rinse, and repeat.

How have you approached controlled vocabulary design for digital assets?

References:

Denton, W. (2003, November). How to Make a Faceted Classification and Put It On the Web. Retrieved from http://www.miskatonic.org/library/facet-web-howto.html

Spiteri, L. (1998). A simplified model for facet analysis: Ranganathan 101. Canadian Journal of Information and Library Science, 23(1/2), 1–30. Retrieved from http://iainstitute.org/en/learn/research/a_simplified_model_for_facet_analysis.php