Smarter Search for South Park - Subdividing Search
(
Page 2 of 2 )
Habib says another reason open source products tend to find their way
into the process is that developers can download them and start working
with them without going through a procurement process. "You can imagine
what the approval process is like for a purchase at a large
corporation. Open source empowers developers come up with solutions,"
he says.
That's what happened with Solr, which one of the developers working on
SouthParkStudios.com began tinkering with on his own initiative, Cohen
says. In that case, the advantage was that Solr made it easier to
subdivide search results according to categories such as the names of
characters featured in that episode, he says.
To achieve those results, the web site must be configured to provide an
XML data feed to Solr that exposes the structure and categories stored
in the site's content management system (CMS). Because MTV's other
major search solution, the Google appliance, relies primarily on
crawling web content, its ability to find search facets is limited to
identifying different file types or extracting categorization
information from HTML meta tags, Cohen says.
So Solr works best with sites such as TheDailyShow.com, which runs off
a single CMS, as opposed to MTV.com, which features layers of
technology built up over decades, Cohen says. Solr is also used on more
complex sites, such as comedycentral.com, as a back end tool for
indexing content but not for serving website search requests.
In addition to helping viewers of TheDailyShow.com find segments
featuring their favorite comic correspondents, Solr feeds a
slider-style timeline user interface widget that sorts through results
by when a segment appeared. Cohen says the faceting feature also makes
it possible to feed search results to multiple international versions
of the MTVMusic site from single index.
The one drawback, Cohen says, is the "tooling" that comes with Solr.
The management and monitoring utilities are not as slick as those that
come with commercial products, he says.
