Solr 101

• 2004: Created by Yonik Seeley (CNET)
• 2006: Apache Software Project
• 2008-9: v1.3-1.4, usage explodes
• 2010: Merged with Lucene
• 2011: v1.4 -> v3.1
• 2014: currently at v4.1

History

• Search engine/server/platform/whatevs
• Written in Java
• Open source, Apache 2.0 license
• Runs as a Jetty or Tomcat servlet
• Provides REST-ish HTTP API (XML or JSON)
• Highly customizable via configuration, plugins
• Also embeddable via EmbeddedSolrServer (although not considered a "best practice")

What is it?

• fulltext searching
• result facets
• term highlighting
• query & index analysis filters
• text extraction (from PDF, Word, etc.)
• Nice admin UI
• easy relevance tuning
• sharding, replication
• NoSQL-ish
• more-like-this, did-you-mean, auto-complete
• nested documents (as of v4.5)
  

Major Features

• Two ways to get content in
  • POST xml or json via HTTP
  • DataImport: import from data source
• Indexing process
  • field-based
  • Defined in schema.xml
  • fieldType defines how field content is processed
  • analysis phase: tokenize, filter, transform
  • storage options: index only, index + content, term frequency, positions, normalizations, vectors
• Inverted Index format: terms -> documents

Indexing

• via HTTP GET: "?q=term1"
• default = Lucene query syntax
  • free-form: term1 term2
  • fielded: foo:term1
  • phrases: "term1 term2"
  • wildcards: "foo:term*"
  • proximity: "term1 term2"~4
  • ranges: [1 TO 1000]
• ExtendedDisMax
  • search across range of fields with varying "boost" factors
  • title:foo^5 fulltext:bar^0.5

Querying

Demo!

• http://lucene.apache.org/solr
• http://www.solrtutorial.com/index.html
• https://wiki.apache.org/solr/FrontPage
• https://slides.com/jamesluker/solr-101

Links