The Faceted Navigation and Search Revolution

Faceted navigation is based on one such pillar of library science, known as faceted classification. This was codified rigorously in the 1930s by S.R. Ranganathan, who is now revered as a founding father of library science, in a book called Colon Classification. As David Weinberger wrote of the master in "Rediscovering Ranganathan" in Forrester Magazine, "Yes, that is the world's worst-ever title for a book, but the system it outlines for classifying books—using categories separated by colons—was revolutionary."

Useful as faceted categories were, in the physical library you still had to ultimately place each book on a single shelf, making it impossible to roam the stacks by the facet of your choice. But mid-century, a librarian named Mortimer Taube took another run at it. Taube's innovation was the Uniterm system, which marries facets and post-coordinate classification to a mechanical index that made the ideal of faceted browsing possible. Finally, the value of faceted classification was unlocked by faceted navigation. In Taube's system, books were cataloged on edge-notched punchcards, where a punched notch indicated the assignment of a facet to that book. To browse for a book, a librarian inserted a rod into a notch in the stack of index cards and lifted, leaving behind just the cards classified by that facet/notch (a variation on the system used light boxes instead of rods to line up notches). Then, he repeated this step with another facet, and so winnowed down to a small set of highly relevant books. Unfortunately, it took considerable training and dexterity to use the system, which means it never made it out from behind the reference desk.

There are some instructive lessons to take away from this brief library history:

(1.) Faceted classification and faceted navigation are complementary—but not the same.

Obvious though it might sound, it took both Ranganathan and Taube to produce a great system. You'll often hear people today use faceted classification interchangeably with faceted navigation, but they don't always come paired. Faceted classification is widespread today; faceted navigation is still too rare. And this isn't just a nit to pick because it leads to our next observation:

(2.) Navigation unlocks the latent value in classification. Almost all content already has some faceted classification or structure that can readily be used for navigation. In fact, rarely have we met a client who didn't think they had unstructured content, yet almost always do we find it to be full of valuable structure. So why don't they recognize its latent value? Until you see it with navigation for the first time, it's not obvious—and once you do, it's indispensable.

(3.) Iteration is the key to browsing. Taube had it right. Contrast these ways of browsing: "I want books on Austrian music history;" or "I want books on music; of those, I want the history ones; of those, I want the ones about Austria."

The former is a search in the ideal world; it assumes that what we're looking for exists, and that we've accurately predicated how the content will be described. It's a pre-coordinate search. The latter corresponds more closely to a real-world search, where we constantly deal with "empty shelves." By taking successive steps along multiple facets, we can winnow down, back up, and avoid dead ends.

(4.) If the user interface isn't simple, the system won't survive.

Taube had the UI wrong, which is why he never made it out from behind the reference desk. His system was good in theory, but everyone must be able to use it, without any training.

Case in Point
While we've been discussing examples of browsing for books in libraries, chances are you're reading this because you want to help your business by helping people reuse documents on intranets, engineering designs in PLM systems, intelligence for Homeland Security or some other kind of electronic enterprise content. In fact, the early adopters of faceted navigation have been enterprises, not libraries. In the past five years, Endeca has launched faceted navigation solutions in 20% of the Fortune 100, and among more of the top 100 online retailers than solutions from all competitors.

So what happens when faceted navigation finally makes it back to Ranganathan and Taube's home turf? In January of 2006, North Carolina State University pioneered the use of faceted navigation on an OPAC, or electronic card catalog, which you can try at www.lib.ncsu.edu/catalog/.

It immediately drew rave reviews from the toughest KM critics out there—the librarians. NCSU themselves called it "a groundbreaking approach...the first of its kind in a library...empowers users to quickly locate the items they're looking for or to explore the multifaceted research collection in depth, exploiting both the software's cutting-edge capabilities and the library's many decades of investment in detailed cataloging and classification." And just a handful of librarian blog excerpts:

• "The best library catalog interface I have come across to date;" 
•  "It's brilliant—faceted browsing and relevance ranking really improve access. It makes the search experience closer to what users expect when searching on the Web;" 
•  "NCSU libraries butcher the pig. Andrew Pace of the NCSU Libraries, who has said that making minor changes to library catalog systems is like putting lipstick on a pig, has unveiled a revolutionary new interface to their catalog;" and
• "The bar has been raised."

Ranganathan and Taube had it right all along, and now without the restrictions of 3-D bookshelves, the power of their principles has come to life.

Guiding Light
So what is the difference between faceted navigation and Guided Navigation? Guided Navigation is the name for Endeca's version of faceted navigation. It's a meeting of classic information science, breakthrough computer science and gritty, real-world software tools and business processes. Guided Navigation wraps search results in a context that shows users how to refine and explore their results, while constantly removing dead ends. With hundreds of commercial deployments now, Endeca has continually built its expertise with leaders like IBM, Home Depot and the DIA back into the best practices reflected in the tools, UIs and APIs that give content owners editorial flexibility and managerial control over their sites, turning faceted navigation into a solution to a business problem.

To learn more about facets and guided navigation, I recommend the first issue of Philip C. Murray's "Barrington Report on Advanced Knowledge Organization and Retrieval," and David Weinberger's "Rediscovering Ranganathan" in Forrester Magazine.

---

Endeca's (www.endeca.com) Guided Navigation®, Search and Analysis solutions help people find, discover and analyze information in ways never before possible. For technology leaders like Walmart.com, NASA, IBM and John Deere, Endeca is creating value by helping their users more effectively find the information they need. And reflecting the broad applicability of these new solutions across the enterprise, Endeca's hundreds of commercial deployments span a wide variety of industries—financial services; media and publishing; manufacturing and distribution; government and retail.