File size: 6,358 Bytes
804f5e1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
# Copyright 2020 The HuggingFace Evaluate Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" seqeval metric. """

import importlib
from typing import List, Optional, Union

import datasets
from seqeval.metrics import accuracy_score, classification_report

import evaluate


_CITATION = """\
@inproceedings{ramshaw-marcus-1995-text,
    title = "Text Chunking using Transformation-Based Learning",
    author = "Ramshaw, Lance  and
      Marcus, Mitch",
    booktitle = "Third Workshop on Very Large Corpora",
    year = "1995",
    url = "https://www.aclweb.org/anthology/W95-0107",
}
@misc{seqeval,
  title={{seqeval}: A Python framework for sequence labeling evaluation},
  url={https://github.com/chakki-works/seqeval},
  note={Software available from https://github.com/chakki-works/seqeval},
  author={Hiroki Nakayama},
  year={2018},
}
"""

_DESCRIPTION = """\
seqeval is a Python framework for sequence labeling evaluation.
seqeval can evaluate the performance of chunking tasks such as named-entity recognition, part-of-speech tagging, semantic role labeling and so on.

This is well-tested by using the Perl script conlleval, which can be used for
measuring the performance of a system that has processed the CoNLL-2000 shared task data.

seqeval supports following formats:
IOB1
IOB2
IOE1
IOE2
IOBES

See the [README.md] file at https://github.com/chakki-works/seqeval for more information.
"""

_KWARGS_DESCRIPTION = """
Produces labelling scores along with its sufficient statistics
from a source against one or more references.

Args:
    predictions: List of List of predicted labels (Estimated targets as returned by a tagger)
    references: List of List of reference labels (Ground truth (correct) target values)
    suffix: True if the IOB prefix is after type, False otherwise. default: False
    scheme: Specify target tagging scheme. Should be one of ["IOB1", "IOB2", "IOE1", "IOE2", "IOBES", "BILOU"].
        default: None
    mode: Whether to count correct entity labels with incorrect I/B tags as true positives or not.
        If you want to only count exact matches, pass mode="strict". default: None.
    sample_weight: Array-like of shape (n_samples,), weights for individual samples. default: None
    zero_division: Which value to substitute as a metric value when encountering zero division. Should be on of 0, 1,
        "warn". "warn" acts as 0, but the warning is raised.

Returns:
    'scores': dict. Summary of the scores for overall and per type
        Overall:
            'accuracy': accuracy,
            'precision': precision,
            'recall': recall,
            'f1': F1 score, also known as balanced F-score or F-measure,
        Per type:
            'precision': precision,
            'recall': recall,
            'f1': F1 score, also known as balanced F-score or F-measure
Examples:

    >>> predictions = [['O', 'O', 'B-MISC', 'I-MISC', 'I-MISC', 'I-MISC', 'O'], ['B-PER', 'I-PER', 'O']]
    >>> references = [['O', 'O', 'O', 'B-MISC', 'I-MISC', 'I-MISC', 'O'], ['B-PER', 'I-PER', 'O']]
    >>> seqeval = evaluate.load("seqeval")
    >>> results = seqeval.compute(predictions=predictions, references=references)
    >>> print(list(results.keys()))
    ['MISC', 'PER', 'overall_precision', 'overall_recall', 'overall_f1', 'overall_accuracy']
    >>> print(results["overall_f1"])
    0.5
    >>> print(results["PER"]["f1"])
    1.0
"""


@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
class Seqeval(evaluate.EvaluationModule):
    def _info(self):
        return evaluate.EvaluationModuleInfo(
            description=_DESCRIPTION,
            citation=_CITATION,
            homepage="https://github.com/chakki-works/seqeval",
            inputs_description=_KWARGS_DESCRIPTION,
            features=datasets.Features(
                {
                    "predictions": datasets.Sequence(datasets.Value("string", id="label"), id="sequence"),
                    "references": datasets.Sequence(datasets.Value("string", id="label"), id="sequence"),
                }
            ),
            codebase_urls=["https://github.com/chakki-works/seqeval"],
            reference_urls=["https://github.com/chakki-works/seqeval"],
        )

    def _compute(
        self,
        predictions,
        references,
        suffix: bool = False,
        scheme: Optional[str] = None,
        mode: Optional[str] = None,
        sample_weight: Optional[List[int]] = None,
        zero_division: Union[str, int] = "warn",
    ):
        if scheme is not None:
            try:
                scheme_module = importlib.import_module("seqeval.scheme")
                scheme = getattr(scheme_module, scheme)
            except AttributeError:
                raise ValueError(f"Scheme should be one of [IOB1, IOB2, IOE1, IOE2, IOBES, BILOU], got {scheme}")
        report = classification_report(
            y_true=references,
            y_pred=predictions,
            suffix=suffix,
            output_dict=True,
            scheme=scheme,
            mode=mode,
            sample_weight=sample_weight,
            zero_division=zero_division,
        )
        report.pop("macro avg")
        report.pop("weighted avg")
        overall_score = report.pop("micro avg")

        scores = {
            type_name: {
                "precision": score["precision"],
                "recall": score["recall"],
                "f1": score["f1-score"],
                "number": score["support"],
            }
            for type_name, score in report.items()
        }
        scores["overall_precision"] = overall_score["precision"]
        scores["overall_recall"] = overall_score["recall"]
        scores["overall_f1"] = overall_score["f1-score"]
        scores["overall_accuracy"] = accuracy_score(y_true=references, y_pred=predictions)

        return scores