Apache Solr

• Search Engine
• Issues with traditional searching
• Performs only sub string matching
• It doesn't understand linguistic variations (buy vs buying are same)
• It doesn't understand synonyms(buying vs purchasing)
• doesn't omit unimportant words(a, the, of..)
• There is no sense of relevancy in results
• It slows down with data increase
• How does Solr solve the traditional search issues
• Solr uses an index that maps contents to documents instead of mapping documents to contents
  
• Inverted Index is at heart of how Search engines work

Characteristics of Search engine

• Text Centric
• Read dominant
• Document Oriented
• Large amount of data
• flexible schema (Relation db is not flexible as every document requires same structure

Install and Startup

• Download jar & extract it
• bin/solr start -p 8983 
• Start Java webserver listenting at 8983 port
• solr is web appln that runs by default in jetty webserver

Solr 

• stores data in documents
• Documents are more flexible than rows in rdbms
• documents can be  hirerachical. rdbms needs different tables & rows
  
• Indexing process

Solr Core

• Single physical index
• directory structure
• server
• solr
• core1
• conf. -> contains managed-schema.xml & solrconfig.xml
  
• data   ->

Solr Document

• Basic unit of solr information
• Json object with key value pairs
• It is similar to dbms table but more flexible
• It can be hierarchical
• It can have array of values
• it can have object as values
• It is denormalized document - all data belonging to an entity is in same document

Indexing Process

• Documented is converted into solr format(json)

Designing Schema

• Create sub-directory on configsets directory 
• copy configuration from _default directory
• 
  

Twitter Search appln

 Elements of text analysis

• analyzer
• tokenizer
• chain of token filers

Faceted Search

•  
  

Other points

• If use case is to write faster then use NO SQL db like Cassandra
•  
•