Faceted Search

This refers to a search model that allows refinement at multiple stages of the search experience based on a predefined or automatically extracted set of facets i.e. attributes of the entities being searched.

For example if the search query is against a list of products that have the following attributes:

  • Name
  • Category
  • Price
  • Color
  • Size
  • Description
  • Image
  • Warranty

Then after or before the original large set of results is returned, the searcher will be given a chance to refine the results by narrowing them down using one or more of the attributes. However, since only attributes with a high probability of sharing values across multiple entities (e.g. products, articles, etc.) are likely to be useful in refinement, administrators usually have the choice of including atributes explicitly for the search refinement. In our example Name, Description and Image are not likely to be the same across different products and would therefore not be suitable for refinement. However, the rest of the attributes can help the searcher reduce the sie of the result set by essentially telling the search application what color, size, category, price or warranty he or she is interested in.

This ia very useful feature due to many reasons:

  1. Since a large percentage of online users are not avid searchers. such users–your potential customers–usually search using high-level generic keywords, even though they might have a clear-enough idea of what low-level attribute values are important to them i.e. a 24-year-old female searching for “black stilettos” who has decided on buying “black leather stilettos” would find the “Leather” value of the “Material” attribute quite useful in quickly eliminating products of no interest from the search results.
  2. Some searchers are not fully decided on the low-level attribute values of what they search for, and could use any help in making up their mind e.g. a 40-year old male searching for “mp3 players” who knows he wants it to hold between 1,000 and 2,000 songs and not in the pink color would find the different non-“pink” values of the “Color” attribute and all values of the “Song Capacity” attribute very useful in making up his mind, hence the term “Guided Navigation.”
  3. Most search algorithms need human administration to produce the best results–and since most e-commerce platforms and website operators ignore this in my experience–faceted search allows users to clear out the noise returned in search results by quickly eliminating unwanted results.

Of course, there are other good reasons to implement a guided navigation feature, depending on the custom needs of each website. There are many vendors that provide “site search” applications with great refinement and guided navigation capabilities. I will post them in the technology category as time allows.

Be Sociable, Share!