"Solr" ~ Erik Hatcher
RubyConf 2007Caching, replication. Very fast, highlighting. Active community. Built on top of Lucene.
Used by CNET, Internet Archive, Netflix, digg.
Lucene is a Java search engine library. Powers Technorati, Monster, Wikipedia, others.
Inverted index under the covers. Takes out the words (“terms”). Documents > Fields > Terms
Includes relevance, does an analysis process which removes stop words (“and”, “or”) and breaks words into stems (removes “-ing”, “-ed”)
Get back Ruby hashes.
solr-ruby is a Ruby DSL for HTTP communication to Solr.
Can set up master Solr server and load balanced read-only searcher slaves.
Relevancy score can be tuned.
The schema consists of dynamicFields. Any searchable fields need to be of type text.
Eric is showing a tool called “Luke” that lets you browse your Lucene index.
ActiveRecord can take a simple hash, so Eric wrote a simple mapper that takes the Solr results in a Ruby Hash and just calls Book.create(mapped_book).
faceting
Needs help with DSL/API guidance, documentation