Search Blog:

Faceted search makes librarians shine (Or, "Why you need a librarian to make your search engine really work.")

By Phil Green, Lucidea COO

I have been working with an early version of the next Presto release. The feature that has me salivating is faceted search (also sometimes referred to as "guided navigation"). Wikipedia defines faceted search as "a technique for accessing information organized according to a
faceted classification system, allowing users to explore a collection of information by applying multiple filters." Basically, if you’ve used an ecommerce site lately, it’s the right- or left-hand navigational aids that help you narrow your search by brand, price, key feature, etc.

Faceted searching is very useful for helping users more easily and quickly find the information that they want – way more efficiently than with full text search alone. This is why almost all ecommerce sites have implemented it. For example, Amazon (King of Ecommerce!) has a particularly good implementation. Ecommerce sites are able to make excellent use of faceted searching because they have facets (or, in database terms, fielded information). For example, I recently purchased an external, USB 3.0, solid state hard drive on Amazon – and getting to the list of available products using faceted search was incredibly easy. I did a quick search, then used the faceted navigation to narrow my results. A few keystrokes later, I was looking at just the drive I wanted to buy!

So why doesn’t Google (King of Search!) use faceted searching? The simple answer is that they don’t have fielded data. They are operating at the level of Web pages and not with well-curated database records (like the ecommerce guys). Which brings me back to my sub-title, and the second reason I like faceted search. Because it makes the work of special librarians shine like no other feature I have ever used. To make faceted search work well, you need two things:

  • First, the database must be carefully constructed to have the fields (facets) that are critical to understanding and describing the items contained in the database. (For example, in ecommerce those would be brand, price, etc.)
  • Second, these fields (facets) must be filled with high-quality content so that guided navigation works well.
In talking to industry consultants, I’ve heard that the ecommerce guys hate guided navigation systems. Why? Because they are such a pain. Full text search is easy. Just feed your information into the search engine and you’re done. Full text search isn’t fussy. No structure to your data? No problem! Low quality inputs? No problem! But to make faceted search work, you have to carefully architect the system and then fill it will high-quality data.

In other words, you often need librarians to help you build a high-quality database.
 Well, guess what? This is what Inmagic clients do every day. When our customers build a knowledge repository, catalog, or archive, they carefully structure the database to have fields (facets) that will truly describe the item – then they use validation lists and controlled vocabulary to fill the database up with high-quality content.

This means that our customers’ databases already meet the requirements for an excellent faceted search experience, which means I can’t wait until our customers have access to the next version of Presto (or Presto for DB/TextWorks) because the results will be amazing. With faceted searching – finding is a lot easier, and the value of your librarian is on display with every search!

Please stay tuned for a follow-up post on the difference between faceted searching and on-the-fly content classifiers (which provide guided navigation, but not true faceted search).



1 comment:

Anonymous said...

This post gets my "Yeah! for the day". Peter, Trimagic Software (Australia)


Related Posts Plugin for WordPress, Blogger...