ethically.we.benchmark.get_data_resource_path() - Code Metrics - Inspection of "Revert "Version 0.0.3 (#14)" (#17)" - ResponsiblyAI/responsibly - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( fe038f...170db5 )

by Shlomi

created 2019-04-10 20:24 UTC

ethically.we.benchmark.get_data_resource_path() A

↳ Parent: ethically.we.benchmark

Complexity

Conditions

Size

Total Lines	4
Code Lines	4

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
cc	1
eloc	4
nop	1
dl	0
loc	4
rs	10
c	0
b	0
f	0

"""
Evaluate words embeedings by standard benchmarks.

Reference: https://github.com/kudkudak/word-embeddings-benchmarks


Word Pairs Tasks

1. The WordSimilarity-353 Test Collection
   http://www.cs.technion.ac.il/~gabr/resources/data/wordsim353/

2. Rubenstein, H., and Goodenough, J. 1965. Contextual correlates of synonymy
   https://www.seas.upenn.edu/~hansens/conceptSim/

3. Stanford Rare Word (RW) Similarity Dataset
   https://nlp.stanford.edu/~lmthang/morphoNLM/

4. The Word Relatedness Mturk-771 Test Collection
   http://www2.mta.ac.il/~gideon/datasets/mturk_771.html

5. The MEN Test Collection
   http://clic.cimec.unitn.it/~elia.bruni/MEN.html

6. SimLex-999
   https://fh295.github.io/simlex.html

7. TR9856
   https://www.research.ibm.com/haifa/dept/vst/files/IBM_Debater_(R)_TR9856.v2.zip


Analogies Tasks

1. Google Analogies (subset of WordRep)
   https://code.google.com/archive/p/word2vec/source

2. MSR - Syntactic Analogies
   http://research.microsoft.com/en-us/projects/rnn/

"""

import os
import warnings

import pandas as pd
from pkg_resources import resource_filename


with warnings.catch_warnings():
    warnings.simplefilter('ignore', category=FutureWarning)


WORD_PAIRS_TASKS = {'WS353': 'wordsim353.tsv',
                    'RG65': 'RG_word.tsv',
                    'RW': 'rw.tsv',
                    'Mturk': 'MTURK-771.tsv',
                    'MEN': 'MEN_dataset_natural_form_full.tsv',
                    'SimLex999': 'SimLex-999.tsv',
                    'TR9856': 'TermRelatednessResults.tsv'}

ANALOGIES_TASKS = {'MSR-syntax': 'MSR-syntax.txt',
                   'Google': 'questions-words.txt'}

PAIR_WORDS_EVALUATION_FIELDS = ['pearson_r', 'pearson_pvalue',
                                'spearman_r', 'spearman_pvalue',
                                'ratio_unkonwn_words']


def get_data_resource_path(filename):
    return resource_filename(__name__, os.path.join('data',
                                                    'benchmark',
                                                    filename))


def prepare_word_pairs_file(src, dst, delimiter='\t'):
    """Transform formats of word pairs files to tsv."""
    df = pd.read_csv(src, header=None, delimiter=delimiter)
    df.loc[:, :2].to_csv(dst, sep=delimiter, index=False, header=False)


def evaluate_word_pairs(model, kwargs_word_pairs=None):
    """
    Evaluate word pairs tasks.

    :param model: Words embedding.
    :param kwargs_word_pairs: Kwargs for
                              evaluate_word_pairs
                              method.
    :type kwargs_word_pairs: dict or None
    :return: DataFrame of evaluation results.
    """

    if kwargs_word_pairs is None:
        kwargs_word_pairs = {}

    results = {}

    for name, filename in WORD_PAIRS_TASKS.items():
        path = get_data_resource_path(filename)
        (pearson,
         spearman,
         ratio_unknown_words) = model.evaluate_word_pairs(path,
                                                          **kwargs_word_pairs)  # pylint: disable=C0301

        results[name] = {'pearson_r': pearson[0],
                         'pearson_pvalue': pearson[1],
                         'spearman_r': spearman.correlation,
                         'spearman_pvalue': spearman.pvalue,
                         'ratio_unkonwn_words': ratio_unknown_words}

    df = (pd.DataFrame(results)
          .reindex(PAIR_WORDS_EVALUATION_FIELDS)
          .transpose()
          .round(3))

    return df


def evaluate_word_analogies(model, kwargs_word_analogies=None):
    """
    Evaluate word analogies tasks.

    :param model: Words embedding.
    :param kwargs_word_analogies: Kwargs for
                                  evaluate_word_analogies
                                  method.
    :type evaluate_word_analogies: dict or None
    :return: DataFrame of evaluation results.
    """

    if kwargs_word_analogies is None:
        kwargs_word_analogies = {}

    results = {}

    for name, filename in ANALOGIES_TASKS.items():
        path = get_data_resource_path(filename)
        overall_score, _ = model.evaluate_word_analogies(path,
                                                         **kwargs_word_analogies)  # pylint: disable=C0301

        results[name] = {'score': overall_score}

        df = (pd.DataFrame(results)
              .transpose()
              .round(3))

    return df



def evaluate_words_embedding(model,
                             kwargs_word_pairs=None,
                             kwargs_word_analogies=None):
    """
    Evaluate word pairs tasks and word analogies tasks.

    :param model: Words embedding.
    :param kwargs_word_pairs: Kwargs fo
                              evaluate_word_pairs
                              method.
    :type kwargs_word_pairs: dict or None
    :param kwargs_word_analogies: Kwargs for
                                  evaluate_word_analogies
                                  method.
    :type evaluate_word_analogies: dict or None
    :return: Tuple of DataFrame for the evaluation results.
    """
    return (evaluate_word_pairs(model, kwargs_word_pairs),
            evaluate_word_analogies(model, kwargs_word_analogies))


1			"""
2			Evaluate words embeedings by standard benchmarks.
3
4			Reference: https://github.com/kudkudak/word-embeddings-benchmarks
5
6
7			Word Pairs Tasks
8
9			1. The WordSimilarity-353 Test Collection
10			http://www.cs.technion.ac.il/~gabr/resources/data/wordsim353/
11
12			2. Rubenstein, H., and Goodenough, J. 1965. Contextual correlates of synonymy
13			https://www.seas.upenn.edu/~hansens/conceptSim/
14
15			3. Stanford Rare Word (RW) Similarity Dataset
16			https://nlp.stanford.edu/~lmthang/morphoNLM/
17
18			4. The Word Relatedness Mturk-771 Test Collection
19			http://www2.mta.ac.il/~gideon/datasets/mturk_771.html
20
21			5. The MEN Test Collection
22			http://clic.cimec.unitn.it/~elia.bruni/MEN.html
23
24			6. SimLex-999
25			https://fh295.github.io/simlex.html
26
27			7. TR9856
28			https://www.research.ibm.com/haifa/dept/vst/files/IBM_Debater_(R)_TR9856.v2.zip
29
30
31			Analogies Tasks
32
33			1. Google Analogies (subset of WordRep)
34			https://code.google.com/archive/p/word2vec/source
35
36			2. MSR - Syntactic Analogies
37			http://research.microsoft.com/en-us/projects/rnn/
38
39			"""
40
41			import os
42			import warnings
43
44			import pandas as pd
45			from pkg_resources import resource_filename
46
47
48			with warnings.catch_warnings():
49			warnings.simplefilter('ignore', category=FutureWarning)
50
51
52			WORD_PAIRS_TASKS = {'WS353': 'wordsim353.tsv',
53			'RG65': 'RG_word.tsv',
54			'RW': 'rw.tsv',
55			'Mturk': 'MTURK-771.tsv',
56			'MEN': 'MEN_dataset_natural_form_full.tsv',
57			'SimLex999': 'SimLex-999.tsv',
58			'TR9856': 'TermRelatednessResults.tsv'}
59
60			ANALOGIES_TASKS = {'MSR-syntax': 'MSR-syntax.txt',
61			'Google': 'questions-words.txt'}
62
63			PAIR_WORDS_EVALUATION_FIELDS = ['pearson_r', 'pearson_pvalue',
64			'spearman_r', 'spearman_pvalue',
65			'ratio_unkonwn_words']
66
67
68			def get_data_resource_path(filename):
69			return resource_filename(__name__, os.path.join('data',
70			'benchmark',
71			filename))
72
73
74			def prepare_word_pairs_file(src, dst, delimiter='\t'):
75			"""Transform formats of word pairs files to tsv."""
76			df = pd.read_csv(src, header=None, delimiter=delimiter)
77			df.loc[:, :2].to_csv(dst, sep=delimiter, index=False, header=False)
78
79
80			def evaluate_word_pairs(model, kwargs_word_pairs=None):
81			"""
82			Evaluate word pairs tasks.
83
84			:param model: Words embedding.
85			:param kwargs_word_pairs: Kwargs for
86			evaluate_word_pairs
87			method.
88			:type kwargs_word_pairs: dict or None
89			:return: DataFrame of evaluation results.
90			"""
91
92			if kwargs_word_pairs is None:
93			kwargs_word_pairs = {}
94
95			results = {}
96
97			for name, filename in WORD_PAIRS_TASKS.items():
98			path = get_data_resource_path(filename)
99			(pearson,
100			spearman,
101			ratio_unknown_words) = model.evaluate_word_pairs(path,
102			**kwargs_word_pairs) # pylint: disable=C0301
103
104			results[name] = {'pearson_r': pearson[0],
105			'pearson_pvalue': pearson[1],
106			'spearman_r': spearman.correlation,
107			'spearman_pvalue': spearman.pvalue,
108			'ratio_unkonwn_words': ratio_unknown_words}
109
110			df = (pd.DataFrame(results)
111			.reindex(PAIR_WORDS_EVALUATION_FIELDS)
112			.transpose()
113			.round(3))
114
115			return df
116
117
118			def evaluate_word_analogies(model, kwargs_word_analogies=None):
119			"""
120			Evaluate word analogies tasks.
121
122			:param model: Words embedding.
123			:param kwargs_word_analogies: Kwargs for
124			evaluate_word_analogies
125			method.
126			:type evaluate_word_analogies: dict or None
127			:return: DataFrame of evaluation results.
128			"""
129
130			if kwargs_word_analogies is None:
131			kwargs_word_analogies = {}
132
133			results = {}
134
135			for name, filename in ANALOGIES_TASKS.items():
136			path = get_data_resource_path(filename)
137			overall_score, _ = model.evaluate_word_analogies(path,
138			**kwargs_word_analogies) # pylint: disable=C0301
139
140			results[name] = {'score': overall_score}
141
142			df = (pd.DataFrame(results)
143			.transpose()
144			.round(3))
145
146			return df
			0 ignored issues – show introduced 2018-09-01 00:53 UTC by Report Bug Copy Issue Report The variable `df` does not seem to be defined in case the `for` loop on line `135` is not entered. Are you sure this can never be the case? Loading history...
147
148
149			def evaluate_words_embedding(model,
150			kwargs_word_pairs=None,
151			kwargs_word_analogies=None):
152			"""
153			Evaluate word pairs tasks and word analogies tasks.
154
155			:param model: Words embedding.
156			:param kwargs_word_pairs: Kwargs fo
157			evaluate_word_pairs
158			method.
159			:type kwargs_word_pairs: dict or None
160			:param kwargs_word_analogies: Kwargs for
161			evaluate_word_analogies
162			method.
163			:type evaluate_word_analogies: dict or None
164			:return: Tuple of DataFrame for the evaluation results.
165			"""
166			return (evaluate_word_pairs(model, kwargs_word_pairs),
167			evaluate_word_analogies(model, kwargs_word_analogies))
168

ResponsiblyAI / responsibly

Push — master ( fe038f...170db5 )

ethically.we.benchmark.get_data_resource_path() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like