Revision history for Lingua-Diversity
0.07 Mon Nov 3 12:15:00 2014
* In L::D::Internals, removed acute accent in comments about Renyi's
entropy.
* In t/03-Lingua-Diversity-Internals, fixed an error in the function
for comparing hashes, which led to spurious failure in Strawberry
Perl.
0.06 Wed Dec 7 22:00:00 2011
* Added Test::Pod::Coverage to build_requires in Build.pl.
* L::D::Utils:
- Fixed version number.
- Fixed bug in _split_tagged_text() (case where none of 'original',
'lemma', and 'tag' was defined in condition).
* L::D::VOCD:
- Set default value of 'min_value' to 0.01 (to work 'per category').
* L::D::Variety:
- Removed the sampling_order attribute, which did not appear to be as
useful as it was confusing. Things now follow the logical
'transform_then_average' pattern. Doc and tests changed accordingly.
* L::D::Subtype:
- Removed the SamplingOrder subtype (no longer used by L::D::Variety).
0.05 Fri Dec 2 11:30:00 2011
* Fixed error in test plan of t/04-Lingua-Diversity-Utils.
* L::D::Subtype:
- Created this module.
- Moved all Moose subtype definitions there.
- Qualified subtypes with L::D namespace (this seems to have been
a problem in v0.04.
- Modified accordingly all modules that used to contain their
own subtype definitions.
- Made a silent prayer for this version to finally pass the tests on
CPAN just like it does on the author's machine.
0.04 Mon Nov 28 23:05:00 2011
* Modified extensive parts of the embedded documentation.
* Added classes L::D::Variety, L::D::SamplingScheme, and L::D::VOCD,
along with corresponding test files.
* Lingua::Diversity (major refactoring):
- Methods measure() and measure_per_category() are not abstract
anymore: they perform the array validation and unit recoding
stuff, and pass the results on to new abstract private method
_measure(). This private method is required to return a
L::D::Result object, which is directly forwarded as the return
value of public method measure() and measure_per_category(). Note
that _measure() has the responsability of handling both the case
where it is passed a single array by measure() and the case where
it is passed two arrays by measure_per_category().
- Subroutines _validate_size() and _prepend_unit_with_category()
have been removed from L::D::Internals and added to this package
(L::D). Tests and exception classes have been removed, moved, or
renamed accordingly.
- Attributes min_num_items and max_num_items (with private getters
and setters) have been added and can be set from within derived
classes if necessary.
- This module now uses L::D::Variety, L::D::MTLD, and L::D::VOCD.
* L::D::MTLD:
- Refactored the code to match the modifications of L::D.
- Fixed bug in _measure(), namely the case of a single partial
factor with a TTR of 1. Now it counts as 1 factor of length 0
(which is not very satisfying but it is hard to come up with a
better alternative).
* L::D::Utils:
- Fixed bug in split_tagged_text() which caused tags to be used in
place of lemmas.
* L::D::Internals:
- Added export tag 'all'.
- Added subroutines _sample_indices(), _count_types(),
_count_frequency(), _shannon_entropy(), _perplexity(),
_renyi_entropy(), and _get_units_per_category() (along with
documentation and tests).
- Moved subroutines _validate_size() and
_prepend_unit_with_category() to the L::D module (along with
documentation and tests).
- Fixed variance precision problem in _get_average().
- Added shortcut in _get_average() for the case where there's only
1 value.
0.03 Sun Nov 13 23:00:00 2011
Fixed a few more errors in the embedded documentation.
Added the possibility to selectively in-/exclude tokens in subroutine
Lingua::Diversity::Utils::split_tagged_text().
Relaxed constraints on attributes 'diversity', 'variance', and 'count'
of Lingua::Diversity::Result objects; they're just plain Nums now.
0.02 Sat Nov 12 21:30:00 2011
Fixed a few typos, errors and glitches in the embedded documentation.
0.01 Sat Nov 12 12:00:00 2011
First version, released on an unsuspecting world.