V3.3.2 2017-05-09
	(no change in this version)

V3.3.1 2017-05-02
	(no change in this version)

V3.3.0 2017-04-07
    - TDQ-13066 SynonymIndexBuilder.getDocument() should always return live records

V3.2.6 2016-12-09
    (no change in this version)

V3.2.5 2016-12-02
	(no change in this version)

V3.2.4 2016-10-20
	TDQ-12143 make classes Serializable to support tSynonymSearch on spark

V3.2.3 2016-09-28
	- TDQ-12143 change org.talend.dataquality.standardization.index.Error class to Serializable to support tSynonymSearch on spark

V3.2.2 2016-09-16
	- move semantic dictionary search API to dataquality-semantic library

V3.2.1 2016-06-27 (with DQ library release 1.5.1)
	(no change in this version)

V3.2.0 2016-05-10
	- rename artifact ID to dataquality-standardization

3.1.0 # 2016-4-27 (for Studio 6.2.0)
	- TDQ-11701 Lucene Index issues 
				* use searcher manager to handle concurrent index access
				* release file handles correctly
				* Removes non-needed dependency on sampling in semantic module.
				* Reuses previously extracted files when found iso. traversing source JAR again and again.
    - TDQ-11440 Add google phone number API.Create an API to validate, format, enrich phone information

3.0.2 2015-12-30
    - move to data-quality repository, change parent pom

3.0.1 # sizhaoliu and fji 2015/11/23
- build standardazation library with new Maven mechanism
- set default maxEdits parameter to 1 instead of 2
- improvements about index matching(TDQ-11143):
    make method getTokensFromAnalyzer() public static and stripping accents
    add 2 new search modes: MATCH_SEMANTIC_DICTIONARY and MATCH_SEMANTIC_KEYWORD
- TDQ-11296 limit the token count for semantic queries to avoid lucene error

3.0.0 # sizhaoliu 2015/10/28
- upgrade lucene library to 4.10.4

2.5.0 # sizhaoliu 2013/09/05 (for V5.4)
- add 6 search modes for synonym library.
  MATCH_ANY MATCH_PARTIAL MATCH_ALL MATCH_EXACT MATCH_ANY_FUZZY MATCH_ALL_FUZZY
- add Matching Threshold option in SynonymIndexSearcher

2.0.2 # sizhaoliu 2012/07/17 
- correct some issues to improve match results of tFirstnameMatch (TDQ-1576) :  
  for standard mode, removed the doc collector and augmented the similarity 
  for fuzzy mode, lowercase the word to assure the prefix is identical to indexed words 

2.0.1 # sizhaoliu 2012/03/15
- Move boost changes from SynonymIndexBuilder to SynonymIndexSearcher, so that we can
  change the boost values without regeneration of the indexes "out of the box".

2.0.0 # sizhaoliu 2012/03/06
- improvements for better scoring with combined query across multiple fields. (related to TDQ-3606)
- do not filter English stop words any more during indexing. (related to TDQ-3330)
- added IndexMigrator to regenerate out-of-the-box indexes.

1.5.19 # sizhaoliu 2011/11/23
- check folder before delete when "initialize" option is chosen in tSynonymOutput component.

1.5.18 # sizhaoliu 2011/05/23
- moved Explainer class to test project.
- included missing Messages classes in the jar.
- added missing trim() when referencing document in update and delete method.

1.5.17 # sizhaoliu 2011/05/19
- added missed trim() for F_WORD during index creation.
- added Explainer class of lucene scoring.  

1.5.16 # ytao 2011/04/15
- change word value of empty record from "" to null
- remove first character '|' of scores
- if input value is empty, create an empty record

1.5.15 # scorreia
- in previous version, the score is not 0 when a field is not match. Hence, a nbMatch field is added to the Output record. 

1.5.14 # scorreia
- SynonymRecordSearch returns a result even when nothing is found. (score is 0 of course). 

1.5.13 # sizhaoliu
- added some other junit tests for new implementation of computeRows
- changed output contents for DOCUMENT_MATCHED

1.5.12 # scorreia
- review implementation of computeRows and add new junit tests.
 
1.5.11 # sizhaoliu
- added trim() before inserting synonyms into index.
- fixed a bug in SynonymRecordSearcher when result counts of different columns have factorization.
- when several documents are matched, set the return value to -1 to avoid insertion by the component (when nbUpdatedDocuments == 0).

1.5.10 # scorreia
- missing classes added to file.

1.5.9 # sizhaoliu
- added information for execution error caused by no segments file
- corrected some user information
- removed the deprecated method setMode()

1.5.8 # scorreia
- code cleansing + javadoc added

1.5.7 # scorreia
- avoid SynonymRecordSearcher to return duplicates
- removed the limit[] argument from the SynonymRecordSearcher.search method

1.5.6 # scorreia sizhaoliu
- add method close after utilisations of SynonymIndexSearcher.
- fixed delete index from file system


1.5.5 # scorreia
- setMode marked as deprecated method (not useful)
- avoid NPE when some synonyms are nulls
- avoid to commit each time we delete a document
- refactor OutputRecord in its own file + add accessors
- search API throws exceptions that can be caught by the component

1.5.4 # ytao
- limit the number of returned results

1.5.3 # scorreia
- optimize synonym index builder and searcher

1.5.2 # ytao modified
- add support to create index path
- add support to return error messages

1.5.1 # ytao modified
- update java doc of deleteDocumentByWord process
- modified insertDocument process,
 remove a clause and change to commit the document immediately
- add java doc for closeIndex process

# Talend standardization library version
1.5.0
