Faceted search is a popular strategy for information exploration to assist the searcher
- Discover relationships or trends between contents.
- To enable dynamic classified browsing.
- Help users who don't know precisely what they can find or what to search for (exploration and discovery).
From the perspective of the information provider they
- Allow one to push searches (navigate them) into specific areas (define top-down narratives).
- Create an aura of relevance.
The primary design goal is to allow users to move through large information spaces in a flexible manner both refining and expanding the queries, while maintaining a consistent representation of the collection's structure and "not getting lost".
Effective faceted search depends on well organized taxonomies. Many organizations, unfortunately, have not done the upfront classification work needed to fully leverage facets for search and guided navigation. The “let’s just use what we have and see what happens” approach may sometimes provide some benefit but nearly always miss their potential and often mislead and navigate users into the wrong direction. Our experience is that the cost of "bad navigation" can be— especially in eCommerce applications— very high. Even with well defined taxonomies we've found that only in homogeneous well-defined settings with shared semantics (men, women, boys, girls, pants, skirts, tops etc.) can they be effective.
Typical implementations of faceted search is via (db intensive) relational database calls. With indexing applied to fields this still tends to limit them to sites with a relatively small number of items (nodes), low update frequency (inserts to indexed fields demand re-indexing) and with comparatively low traffic.
IB handles things differently. Information is de-coupled from the RDMS and off-loaded. Faceted search is implemented in IB via cacheable search calls including:
- Field/Path faceting: the counts for all terms, or just the top terms in any given field or path.
- Path Fragement Faceting: Like field/path faceting but is a relative path (in the document tree) via the field/path instance where a given query found a match.
- Query faceting: the number of documents in the current search results that also match the given query.
- Date faceting: the number of documents that fall within certain date ranges.
- Data faceting: the number of documents that fall with a numerical range in a specific numerical field, geospatial bounding box etc.
IB also features a unique feature called magnetism. It allows (set via per-index parameters) to allow items in a result list to "attract" to one another (or repel) allowing them to fall into segmented clusters.
Like Search and Scan, Faceted Search its offloaded from the application front end (CMS etc.) and provided as services via the server.
