ArneBinder commited on
Commit
1ef2db4
1 Parent(s): c5a482a

initial release from https://github.com/ArneBinder/pie-datasets/pull/106

Browse files
Files changed (3) hide show
  1. README.md +50 -0
  2. drugprot.py +154 -0
  3. requirements.txt +1 -0
README.md ADDED
@@ -0,0 +1,50 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # PIE Dataset Card for "DrugProt"
2
+
3
+ This is a [PyTorch-IE](https://github.com/ChristophAlt/pytorch-ie) wrapper for the
4
+ [DrugProt Huggingface dataset loading script](https://huggingface.co/datasets/bigbio/drugprot).
5
+
6
+ ## Data Schema
7
+
8
+ There are two versions of the dataset supported, `drugprot_source` and `drugprot_bigbio_kb`.
9
+
10
+ #### `DrugprotDocument` for `drugprot_source`
11
+
12
+ defines following fields:
13
+
14
+ - `text` (str)
15
+ - `id` (str, optional)
16
+ - `metadata` (dictionary, optional)
17
+ - `title` (str, optional)
18
+ - `abstract` (str, optional)
19
+
20
+ and the following annotation layers:
21
+
22
+ - `entities` (annotation type: `LabeledSpan`, target: `text`)
23
+ - `relations` (annotation type: `BinaryRelation`, target: `entities`)
24
+
25
+ #### `DrugprotBigbioDocument` for `drugprot_bigbio_kb`
26
+
27
+ defines following fields:
28
+
29
+ - `text` (str)
30
+ - `id` (str, optional)
31
+ - `metadata` (dictionary, optional)
32
+
33
+ and the following annotation layers:
34
+
35
+ - `passages` (annotation type: `LabeledSpan`, target: `text`)
36
+ - `entities` (annotation type: `LabeledSpan`, target: `text`)
37
+ - `relations` (annotation type: `BinaryRelation`, target: `entities`)
38
+
39
+ See [here](https://github.com/ArneBinder/pie-modules/blob/main/src/pie_modules/annotations.py) for the annotation
40
+ type definitions.
41
+
42
+ ## Document Converters
43
+
44
+ The dataset provides predefined document converters for the following target document types:
45
+
46
+ - `pie_modules.documents.TextDocumentWithLabeledSpansAndBinaryRelations` for `DrugprotDocument`
47
+ - `pie_modules.documents.TextDocumentWithLabeledSpansBinaryRelationsAndLabeledPartitions` for `DrugprotBigbioDocument`
48
+
49
+ See [here](https://github.com/ArneBinder/pie-modules/blob/main/src/pie_modules/documents.py) for the document type
50
+ definitions.
drugprot.py ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ from dataclasses import dataclass
2
+ from typing import Any, Dict, Optional, Union
3
+
4
+ import datasets
5
+ from pie_modules.annotations import BinaryRelation, LabeledSpan
6
+ from pie_modules.documents import (
7
+ AnnotationLayer,
8
+ TextBasedDocument,
9
+ TextDocumentWithLabeledSpansAndBinaryRelations,
10
+ TextDocumentWithLabeledSpansBinaryRelationsAndLabeledPartitions,
11
+ annotation_field,
12
+ )
13
+
14
+ from pie_datasets import GeneratorBasedBuilder
15
+
16
+
17
+ @dataclass
18
+ class DrugprotDocument(TextBasedDocument):
19
+ title: Optional[str] = None
20
+ abstract: Optional[str] = None
21
+ entities: AnnotationLayer[LabeledSpan] = annotation_field(target="text")
22
+ relations: AnnotationLayer[BinaryRelation] = annotation_field(target="entities")
23
+
24
+
25
+ @dataclass
26
+ class DrugprotBigbioDocument(TextBasedDocument):
27
+ passages: AnnotationLayer[LabeledSpan] = annotation_field(target="text")
28
+ entities: AnnotationLayer[LabeledSpan] = annotation_field(target="text")
29
+ relations: AnnotationLayer[BinaryRelation] = annotation_field(target="entities")
30
+
31
+
32
+ def example2drugprot(example: Dict[str, Any]) -> DrugprotDocument:
33
+ metadata = {"entity_ids": []}
34
+ id2labeled_span: Dict[str, LabeledSpan] = {}
35
+
36
+ document = DrugprotDocument(
37
+ text=example["text"],
38
+ title=example["title"],
39
+ abstract=example["abstract"],
40
+ id=example["document_id"],
41
+ metadata=metadata,
42
+ )
43
+ for span in example["entities"]:
44
+ labeled_span = LabeledSpan(
45
+ start=span["offset"][0],
46
+ end=span["offset"][1],
47
+ label=span["type"],
48
+ )
49
+ document.entities.append(labeled_span)
50
+ document.metadata["entity_ids"].append(span["id"])
51
+ id2labeled_span[span["id"]] = labeled_span
52
+ for relation in example["relations"]:
53
+ document.relations.append(
54
+ BinaryRelation(
55
+ head=id2labeled_span[relation["arg1_id"]],
56
+ tail=id2labeled_span[relation["arg2_id"]],
57
+ label=relation["type"],
58
+ )
59
+ )
60
+ return document
61
+
62
+
63
+ def example2drugprot_bigbio(example: Dict[str, Any]) -> DrugprotBigbioDocument:
64
+ text = " ".join([" ".join(passage["text"]) for passage in example["passages"]])
65
+ doc_id = example["document_id"]
66
+ metadata = {"entity_ids": []}
67
+ id2labeled_span: Dict[str, LabeledSpan] = {}
68
+
69
+ document = DrugprotBigbioDocument(
70
+ text=text,
71
+ id=doc_id,
72
+ metadata=metadata,
73
+ )
74
+ for passage in example["passages"]:
75
+ document.passages.append(
76
+ LabeledSpan(
77
+ start=passage["offsets"][0][0],
78
+ end=passage["offsets"][0][1],
79
+ label=passage["type"],
80
+ )
81
+ )
82
+ # We sort labels and relation to always have an deterministic order for testing purposes.
83
+ for span in example["entities"]:
84
+ labeled_span = LabeledSpan(
85
+ start=span["offsets"][0][0],
86
+ end=span["offsets"][0][1],
87
+ label=span["type"],
88
+ )
89
+ document.entities.append(labeled_span)
90
+ document.metadata["entity_ids"].append(span["id"])
91
+ id2labeled_span[span["id"]] = labeled_span
92
+ for relation in example["relations"]:
93
+ document.relations.append(
94
+ BinaryRelation(
95
+ head=id2labeled_span[relation["arg1_id"]],
96
+ tail=id2labeled_span[relation["arg2_id"]],
97
+ label=relation["type"],
98
+ )
99
+ )
100
+ return document
101
+
102
+
103
+ class Drugprot(GeneratorBasedBuilder):
104
+ DOCUMENT_TYPES = {
105
+ "drugprot_source": DrugprotDocument,
106
+ "drugprot_bigbio_kb": DrugprotBigbioDocument,
107
+ }
108
+
109
+ BASE_DATASET_PATH = "bigbio/drugprot"
110
+ BASE_DATASET_REVISION = "38ff03d68347aaf694e598c50cb164191f50f61c"
111
+
112
+ BUILDER_CONFIGS = [
113
+ datasets.BuilderConfig(
114
+ name="drugprot_source",
115
+ version=datasets.Version("1.0.2"),
116
+ description="DrugProt source version",
117
+ ),
118
+ datasets.BuilderConfig(
119
+ name="drugprot_bigbio_kb",
120
+ version=datasets.Version("1.0.0"),
121
+ description="DrugProt BigBio version",
122
+ ),
123
+ ]
124
+
125
+ @property
126
+ def document_converters(self):
127
+ if self.config.name == "drugprot_source":
128
+ return {
129
+ TextDocumentWithLabeledSpansAndBinaryRelations: {
130
+ "entities": "labeled_spans",
131
+ "relations": "binary_relations",
132
+ }
133
+ }
134
+ elif self.config.name == "drugprot_bigbio_kb":
135
+ return {
136
+ TextDocumentWithLabeledSpansBinaryRelationsAndLabeledPartitions: {
137
+ "passages": "labeled_partitions",
138
+ "entities": "labeled_spans",
139
+ "relations": "binary_relations",
140
+ }
141
+ }
142
+ else:
143
+ raise ValueError(f"Unknown dataset name: {self.config.name}")
144
+
145
+ def _generate_document(
146
+ self,
147
+ example: Dict[str, Any],
148
+ ) -> Union[DrugprotDocument, DrugprotBigbioDocument]:
149
+ if self.config.name == "drugprot_source":
150
+ return example2drugprot(example)
151
+ elif self.config.name == "drugprot_bigbio_kb":
152
+ return example2drugprot_bigbio(example)
153
+ else:
154
+ raise ValueError(f"Unknown dataset config name: {self.config.name}")
requirements.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ pie-datasets>=0.9.0,<0.10.0