Search Engine Information / MediaWiki Search Extension

Platform: Unix, Linux , Windows, MacOS X (written in C++)
Price: free open source under GPL.

Features

• Indexes via SphinxSE direct connection to MySQL or PostgreSQL
• Accepts data in XML format from using the Sphinx API
• Note that this package does not include a robot spider for traversing web sites.
• High-speed indexing, up to 10 MB/sec on fast CPU
• Can store up to 100 GB text / 100 million documents on a single CPU
• Indexes any number of fields and metadata tags, can be changed on the fly
• Stemming for English and Russian words
• Simple search modes, "match all", "match any", "match phrase"
  
• Also supports Boolean expressions (using &, |, -, and parentheses)
• Field searching, wildcarding, and proximity operators
• Fast searching (under 0.1 second on 2-4 GB index)
• Stemming query processing English and Russian natively, and with libstemmer for French, Spanish, Portuguese, Italian, German, Dutch, Swedish, Norwegian, Danish, Finnish, Hungarian
• Relevance based on BM25, (a flavor of TF:IDF) and phrase-based ranking.
• Sort by relevance, metadata in ascending or descending order, time range, or a SQL-like expression
• Group results by day, week, month, year, or an attribute value, and show only the best match per group.
• Can adjust relevance according to fields (even at runtime)
• Options for faceted metadata display
• XML results interface for custom integration
• Can replace SQL WHERE, ORDER BY, and GROUP BY query elements
• APIs for PHP, Python, Java, Perl, and Ruby
• Can utilize many CPU cores
• Distributed search - multiple search engines can search segmented indexes on multiple machines.
• Scales to at least 1.2 billion records, 30 million searches per day on a 15-server installation
• Source code and binaries available
• Administration by config file

Articles & Reviews

• Sphinx - text search the Pirate Bay wayTheRegister.com: May 8 2009, by Ted Dziuba.
  Describes how Sphinx replaced the MySQL search at Craigslist, cutting the server requirements in half, and runs the Pirate Bay torrent search engine. Includes some feature info and comparison with Solr. There are two very happy Sphinx users in the comments, just ignore the MySQL vs. PHP argument.
  
• Installing Sphinx & Ultrasphinx on Windows Cordinc Blog: March 26, 2008 by Charles
  Notes about how to install Sphinx and make it work with Ruby on Rails.
  
• Searching database content with Sphinx Linux.com: August 24, 2007 by Frederico Kereki
  Nice overview and introduction to Sphinx.
  
• Build a custom search engine with PHP IBM developerworks: July 31, 2007 by Martin Streicher
  
  A hands-on how-to article using an example of auto-body parts, a complex MySQL table to store the data, and Sphinx
  to search, especially for "OR" searches and complex queries.
• Sphinx: Going Beyond full text search MySQL Performance Blog: July 23, 2007 by peter
  Describes using Sphinx to do hundreds of thousands of queries per day against a billion forum posts at boardreader.com
• PHP Search Engine Showdown ONLamp.com : April 03, 2007 , by Michael Douma
  
  First page has a nice overview of site search for database-backed sites, second page has quick reviews of iSearch (1 star), MnogoSearch (3 stars), PHPDig (2 stars), RiSearch (1 star), Sphider (3 stars), Sphinx( 3 stars). and TSEP
  (two stars).

Examples

• Apartment Therapy - interior and exterior design, social networking site
• eScience News - popular science news aggregator
• Helsingborgs Dagblad - a large newspaper in Sweden.
• Absorbent Printing - custom promotional and corporate gift printing
• New World Encyclopedia - an online encyclopedia