Faceted searching works well when your content is well organized and well cataloged. What exactly does that mean, and what are the benefits of faceted searching?

Let me illustrate the benefits of facets applied to well organized, well cataloged content, using a collection of books as an example. Facets based on the Author field aren’t normally very useful. Author is most often a unique field. Therefore, the Author facet is simply a list of all the Authors, which is not very helpful for refining your search results. On the other hand, if the books are well cataloged, and have been classified by Type as well as Subject(s), using a controlled vocabulary, the Type facet will show you a smaller list of Types of books (e.g. Biography, Travel, Fiction-Mystery, etc.) and the Subject facet will show you a smaller list of key subjects (Medical Research, Medieval England, Metallurgy, etc.). These shorter lists help users efficiently navigate search results so that they can find the book they seek, such as a travel guide to medieval English sites.

Faceted search definitely offers many benefits. For example, one of the great things about facets/fields is that they are domain specific, so librarians and subject specialists can build specific taxonomies for each domain - and offer very powerful access routes to content. By the way, Amazon is a great example of faceted search capabilities because their content is structured and organized (e.g. External hard drive; USB interface; 1.5 Gigabytes), whereas Google doesn’t do faceted searching as they don't use fields and must rely on full text searching – which can make it harder to quickly find specific content.

You might be wondering how best to develop the controlled vocabularies (e.g. “taxonomies”) for each field. This can be done in several ways:

• The old fashioned way – speak to users, gather information from domain experts, and build a list of terms commonly used for the field you want to facet
• User behavior based – gather search statistics, review social tags, etc. and build the controlled vocabulary based on this input
• Use a combination of the above.

Of course, the next question is “How can you apply the newly developed controlled vocabulary to the collection?” That is definitely an important issue, and again, it is often a combination of manual work and effective automation. We’ll tackle that topic in a future post, so please stay tuned.