Update README.md
Browse files
README.md
CHANGED
@@ -7,16 +7,16 @@ dataset_info:
|
|
7 |
dtype: string
|
8 |
splits:
|
9 |
- name: train
|
10 |
-
num_bytes: 112840658
|
11 |
num_examples: 109
|
12 |
- name: val
|
13 |
-
num_bytes: 10941585
|
14 |
num_examples: 10
|
15 |
- name: test
|
16 |
-
num_bytes: 27477418
|
17 |
num_examples: 26
|
18 |
download_size: 150586805
|
19 |
-
dataset_size: 151259661
|
20 |
configs:
|
21 |
- config_name: default
|
22 |
data_files:
|
@@ -26,4 +26,48 @@ configs:
|
|
26 |
path: data/val-*
|
27 |
- split: test
|
28 |
path: data/test-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
dtype: string
|
8 |
splits:
|
9 |
- name: train
|
10 |
+
num_bytes: 112840658
|
11 |
num_examples: 109
|
12 |
- name: val
|
13 |
+
num_bytes: 10941585
|
14 |
num_examples: 10
|
15 |
- name: test
|
16 |
+
num_bytes: 27477418
|
17 |
num_examples: 26
|
18 |
download_size: 150586805
|
19 |
+
dataset_size: 151259661
|
20 |
configs:
|
21 |
- config_name: default
|
22 |
data_files:
|
|
|
26 |
path: data/val-*
|
27 |
- split: test
|
28 |
path: data/test-*
|
29 |
+
license: apache-2.0
|
30 |
+
task_categories:
|
31 |
+
- text-generation
|
32 |
+
language:
|
33 |
+
- pl
|
34 |
+
tags:
|
35 |
+
- legal
|
36 |
+
pretty_name: Polish Insurance Terms Structuring
|
37 |
+
size_categories:
|
38 |
+
- n<1K
|
39 |
---
|
40 |
+
|
41 |
+
|
42 |
+
A dataset for the task of document structuring (parsing) Polish legal documents with nested lists.
|
43 |
+
|
44 |
+
|
45 |
+
The dataset contains 2 columns:
|
46 |
+
- `image`: pdf pages converted to 1080x1440 images,
|
47 |
+
- `gt_json`: a json containing a list of detected objects under `gt_parse` key.
|
48 |
+
|
49 |
+
|
50 |
+
The detected objects in the `gt_json` column are specified as follows:
|
51 |
+
|
52 |
+
| Object Type | Keys | Description |
|
53 |
+
|-------------|------|-------------|
|
54 |
+
| Heading | content | Represents a heading-like loose block of text. |
|
55 |
+
| List | title, items | A collection of items, which can be an element, an overflowing element, or another list. |
|
56 |
+
| Element | content | A basic unit of content within a list (along with the list prefix, *e.g.* "1.") or a loose block of text. |
|
57 |
+
| Overflowing Element | content | Content that continues from the main list or element, but isn't part of the standard list flow. |
|
58 |
+
| Element Continuation | content | Content that continues from an element on the previous page. |
|
59 |
+
| List Continuation | items | List continuation from the previous page; items can be an element, an overflowing element, or another list. |
|
60 |
+
|
61 |
+
|
62 |
+
All images were obtained from Polish insurance companies official websites. All sources are provided below:
|
63 |
+
|
64 |
+
- [01-OWU-Compensa-Komunikacja](https://www.compensa.pl/app/owu/fileadmin/user_upload/OWU/OWU_Compensa_Komunikacja.pdf)
|
65 |
+
- [03-OWU-PZU-Wojazer](https://www.pzu.pl/_fileserver/item/1538261)
|
66 |
+
- [04-OWU-PZU-Jachty](https://www.pzu.pl/_fileserver/item/1513488)
|
67 |
+
- [05-OWU-PZU-Auto](https://www.pzu.pl/_fileserver/item/1553025)
|
68 |
+
- [06-OWU-Compensa-Rodzina](https://www.compensa.pl/app/uploads/2024/06/OWU-Compensa-Rodzina-IV-wersja-jednolita-z-aneksem-nr-3_dla-umow-zawieranych-od-15.06.23_nowy-kapital.pdf)
|
69 |
+
- [07-OWU-Compensa-Zdrowie](https://www.compensa.pl/app/uploads/2022/11/OWU-Compensa-Zdrowie-270-2022-1.pdf)
|
70 |
+
- [08-OWU-Compensa-NNW](https://www.compensa.pl/app/uploads/2024/06/OWU_NNW-2024.pdf)
|
71 |
+
- [10-OWU-PZU-NNW](https://www.pzu.pl/_fileserver/item/1513529)
|
72 |
+
- [11-OWU-ErgoHestia-Zycie](https://cdn.bsbox.pl/files/hestia/NTc7MDA_/90a42ad9b4f9165e24d889ebba6c0afc_documents-c10b63ff-85aa-4f33-975d-3b06b14caea8.pdf)
|
73 |
+
- [12-OWU-ErgoHestia-Motopakiet](https://cdn.bsbox.pl/files/hestia/ODQ7MDA_/c11b39b09cfcf3f33eb4b5595e7dd088_documents-2cc88278-bd70-49d1-8d8b-0e72101d2e72.pdf)
|