ArneBinder commited on
Commit
524ffec
1 Parent(s): 6b69296

from https://github.com/ArneBinder/pie-datasets/pull/140

Browse files
Files changed (3) hide show
  1. README.md +34 -0
  2. requirements.txt +1 -0
  3. tbga.py +119 -0
README.md ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # PIE Dataset Card for "TBGA"
2
+
3
+ This is a [PyTorch-IE](https://github.com/ChristophAlt/pytorch-ie) wrapper for the
4
+ [TBGA Huggingface dataset loading script](https://huggingface.co/datasets/DFKI-SLT/tbga).
5
+
6
+ ## Data Schema
7
+
8
+ The document type for this dataset is `TbgaDocument` which defines the following data fields:
9
+
10
+ - `text` (str)
11
+
12
+ and the following annotation layers:
13
+
14
+ - `entities` (annotation type: `SpanWithIdAndName`, target: `text`)
15
+ - `relations` (annotation type: `BinaryRelation`, target: `entities`)
16
+
17
+ `SpanWithIdAndName` is a custom annotation type that extends typical `Span` with the following data fields:
18
+
19
+ - `id` (str, for entity identification)
20
+ - `name` (str, entity string between span start and end)
21
+
22
+ See [here](https://github.com/ArneBinder/pie-modules/blob/main/src/pie_modules/annotations.py) and
23
+ [here](https://github.com/ChristophAlt/pytorch-ie/blob/main/src/pytorch_ie/annotations.py) for the annotation
24
+ type definitions.
25
+
26
+ ## Document Converters
27
+
28
+ The dataset provides predefined document converters for the following target document types:
29
+
30
+ - `pie_modules.documents.TextDocumentWithLabeledSpansAndBinaryRelations`
31
+
32
+ See [here](https://github.com/ArneBinder/pie-modules/blob/main/src/pie_modules/documents.py) and
33
+ [here](https://github.com/ChristophAlt/pytorch-ie/blob/main/src/pytorch_ie/documents.py) for the document type
34
+ definitions.
requirements.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ pie-datasets>=0.6.0,<0.11.0
tbga.py ADDED
@@ -0,0 +1,119 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import dataclasses
2
+ from typing import Any
3
+
4
+ import datasets
5
+ from pytorch_ie import AnnotationLayer, annotation_field
6
+ from pytorch_ie.annotations import BinaryRelation, LabeledSpan, Span
7
+ from pytorch_ie.documents import (
8
+ TextBasedDocument,
9
+ TextDocumentWithLabeledSpansAndBinaryRelations,
10
+ )
11
+
12
+ from pie_datasets import ArrowBasedBuilder, GeneratorBasedBuilder
13
+
14
+
15
+ @dataclasses.dataclass(frozen=True)
16
+ class SpanWithIdAndName(Span):
17
+ id: str
18
+ name: str
19
+
20
+ def resolve(self) -> Any:
21
+ return self.id, self.name, super().resolve()
22
+
23
+
24
+ @dataclasses.dataclass
25
+ class TbgaDocument(TextBasedDocument):
26
+ entities: AnnotationLayer[SpanWithIdAndName] = annotation_field(target="text")
27
+ relations: AnnotationLayer[BinaryRelation] = annotation_field(target="entities")
28
+
29
+
30
+ def example_to_document(example) -> TbgaDocument:
31
+ document = TbgaDocument(text=example["text"])
32
+ head = SpanWithIdAndName(
33
+ # this is due to the original dataset having an integer id but string is required
34
+ id=str(example["h"]["id"]),
35
+ name=example["h"]["name"],
36
+ start=example["h"]["pos"][0],
37
+ end=example["h"]["pos"][0] + example["h"]["pos"][1], # end is start + length
38
+ )
39
+ tail = SpanWithIdAndName(
40
+ id=example["t"]["id"],
41
+ name=example["t"]["name"],
42
+ start=example["t"]["pos"][0],
43
+ end=example["t"]["pos"][0] + example["t"]["pos"][1], # end is start + length
44
+ )
45
+ document.entities.extend([head, tail])
46
+
47
+ relation = BinaryRelation(head=head, tail=tail, label=example["relation"])
48
+ document.relations.append(relation)
49
+ return document
50
+
51
+
52
+ def document_to_example(document):
53
+ head = document.entities[0]
54
+ tail = document.entities[1]
55
+ return {
56
+ "text": document.text,
57
+ "relation": document.relations[0].label,
58
+ "h": {"id": int(head.id), "name": head.name, "pos": [head.start, head.end - head.start]},
59
+ "t": {"id": tail.id, "name": tail.name, "pos": [tail.start, tail.end - tail.start]},
60
+ }
61
+
62
+
63
+ def convert_to_text_document_with_labeled_spans_and_binary_relations(
64
+ document: TbgaDocument,
65
+ ) -> TextDocumentWithLabeledSpansAndBinaryRelations:
66
+ text_document = TextDocumentWithLabeledSpansAndBinaryRelations(text=document.text)
67
+ old2new_spans = {}
68
+ ids = []
69
+ names = []
70
+
71
+ for entity in document.entities: # in our case two entities (head and tail)
72
+ # create LabeledSpan and append
73
+ labeled_span = LabeledSpan(start=entity.start, end=entity.end, label="ENTITY")
74
+ text_document.labeled_spans.append(labeled_span)
75
+
76
+ # Map the original entity to the new labeled span
77
+ old2new_spans[entity] = labeled_span
78
+
79
+ ids.append(entity.id)
80
+ names.append(entity.name)
81
+
82
+ if len(document.relations) != 1: # one relation between two entities
83
+ raise ValueError(f"Expected exactly one relation, got {len(document.relations)}")
84
+ old_rel = document.relations[0]
85
+
86
+ # create BinaryRelation and append
87
+ rel = BinaryRelation(
88
+ head=old2new_spans[old_rel.head],
89
+ tail=old2new_spans[old_rel.tail],
90
+ label=old_rel.label,
91
+ )
92
+ text_document.binary_relations.append(rel)
93
+ text_document.metadata["entity_ids"] = ids
94
+ text_document.metadata["entity_names"] = names
95
+
96
+ return text_document
97
+
98
+
99
+ class Tbga(ArrowBasedBuilder):
100
+ DOCUMENT_TYPE = TbgaDocument
101
+ BASE_DATASET_PATH = "DFKI-SLT/tbga"
102
+ BASE_DATASET_REVISION = "78575b79aa1c6ff7712bfa0f0eb0e3d01d80e9bc"
103
+
104
+ BUILDER_CONFIGS = [
105
+ datasets.BuilderConfig(
106
+ version=datasets.Version("1.0.0"),
107
+ description="TBGA dataset",
108
+ )
109
+ ]
110
+
111
+ DOCUMENT_CONVERTERS = {
112
+ TextDocumentWithLabeledSpansAndBinaryRelations: convert_to_text_document_with_labeled_spans_and_binary_relations
113
+ }
114
+
115
+ def _generate_document(self, example, **kwargs):
116
+ return example_to_document(example)
117
+
118
+ def _generate_example(self, document, **kwargs):
119
+ return document_to_example(document)