pipeline/nltk/test/meteor.doctest

.. Copyright (C) 2001-2023 NLTK Project
.. For license information, see LICENSE.TXT

.. -*- coding: utf-8 -*-

=============
METEOR tests
=============

No Alignment test
------------------

    >>> from nltk.translate import meteor
    >>> from nltk import word_tokenize

If the candidate has no alignment to any of the references, the METEOR score is 0.

    >>> round(meteor(
    ...     [word_tokenize('The candidate has no alignment to any of the references')],
    ...     word_tokenize('John loves Mary')
    ... ), 4)
    0.0

Tests based on wikipedia examples
---------------------------------

Testing on `wikipedia examples <https://en.wikipedia.org/wiki/METEOR#Examples>`_

    >>> same_res = round(meteor(
    ...       [word_tokenize('The cat sat on the mat')],
    ...       word_tokenize('The cat sat on the mat')
    ...       ), 4)
    >>> abs(same_res - 0.9977) < 1e-2
    True

    >>> meteor(
    ...       [word_tokenize('The cat sat on the mat')],
    ...       word_tokenize('on the mat sat the cat')
    ...       )
    0.5

    >>> round(meteor(
    ...       [word_tokenize('The cat sat on the mat')],
    ...       word_tokenize('The cat was sat on the mat')
    ...       ), 4)
    0.9654

Test corresponding to issue #2751, where METEOR score > 1

    >>> round(meteor(
    ...       [word_tokenize('create or update a vm set')],
    ...       word_tokenize('creates or updates a virtual machine scale set')
    ...       ), 4)
    0.7806