Faceted Search

A mechanism to help users filter a search result set to find the most appropriate articles.  The average length of a search query is 1.5 words, so unsurprisingly the results returned by a free-text engine will include the information the user was expecting and a lot of information they weren’t.

How does faceted search work?

Facets are the topic areas a user might expect to use to further filter and refine the result set.  This is most obvious on online retail sites where the facets would include brand, size, color, etc.  For information portals the facets could be Subject, Geography, Author, Format or Department.

Each Facet will show a list of appropriate terms sourced from a controlled vocabulary or taxonomy (and managed in a tool like Ontology Manager).  Selecting a term should filter the results so a sub-set of the original results are shown.  The mechanism for filtering varies.  At its simplest the facet term is simply added to the original search query (free-text results for “users query” AND “filter term”).  A more powerful filter is available when the content has been accurately tagged (manually or by a tool like Semaphore Classification Server).  The filter can then be all the items from the original result set that have a metadata value of the facet value.

Sometimes displaying the facets in a hierarchy provides more context.  A service like Semantic Enhancement Server can deliver the taxonomy tree information (and indeed drop of nodes where there are no matching content items).

The level of facet support differs across search platforms.  Google Search Appliance and Apache Solr provide the basic indexing structures which can be populated and used by Semaphore to deliver a compelling search user experience.