CHANGES.txt

1.3 (2022-02-14)
------------------

- remove normalization; it does not belong here
- switch to ISO release date format

1.2.4 (11.2. 2022)
------------------

- small re fix
- fix README

1.2.3 (10.2. 2022)
------------------

- have bestmatch return match quality also

1.2.1 (6.2. 2017)
-----------------

- constrain acceptable multiple match reduction by alternative
  length (difference from term length)

1.2 (2.2. 2017)
---------------

- add special handling for cases where there's a continuous-
  substring match but it's not enough to produce a sufficiently
  high-quality match (see code for specific algorithm)

- now has 99,7% correct match rate with real-life
  test data consisting of about 1000 unique samples

1.1 (1.2. 2017)
---------------

- switch to assuming that data passed in is already normalized
  to requirements, whatever they are (keep turning everything
  into lowercase though); change test data and test data
  generation accordingly (things run a bit faster now)
- improved detection; common substring matches are now 
  distinguished by length, and the normalized match whose
  length is closest to term in character length, is chosen as
  the match (closes #4)
- cleanup and refactoring (new util module)
- improved tests

1.0 (31.3. 2016)
----------------

- assert variations passed to `closestvariation` is nonempty
- add support for handling multiple matches, with tests

0.9 (14.9. 2015)
----------------

- add support for Damerau-Levenshtein algorithm
- add support for detecting multiple equally good matches
- add support for optionally disqualifying terms from being
  used as best match results
- add support for generating additional variations from dashed
  and dotted names to improve matching
- make sure using unicode throughout works
- return more info in matching result
- add draft Sift4 matching algorithm support
- tests improvements & refactoring
- add test data for future testing of ambiguous matches

0.8 (9.9. 2015)
---------------

- use cleanco 1.3 unicode mode

0.7 (2.9. 2015)
---------------

- support Jaccard, Levensthein & Sorensen as
  matching algorithms

0.6 (2.9. 2015)
---------------

- normalize also the name, not just variations

0.5 (1.9. 2015)
---------------

- store match info in NoMatchError

0.4 (27.8. 2015)
-----------------

- raise a custom NoMatchError upon failure