Update README.md
Browse files
README.md
CHANGED
@@ -9,10 +9,23 @@ size_categories:
|
|
9 |
- 1M<n<10M
|
10 |
---
|
11 |
|
12 |
-
|
13 |
|
|
|
14 |
|
15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
|
17 |
Please reference the following if you make use of this dataset:
|
18 |
|
|
|
9 |
- 1M<n<10M
|
10 |
---
|
11 |
|
12 |
+
## Data Introduction
|
13 |
|
14 |
+
Over 1.5 Million synthetically generated ground-truth/[OCR](https://en.wikipedia.org/wiki/Optical_character_recognition) pairs for post correction tasks from our paper "[Large Synthetic Data from the ar𝜒iv for OCR Post Correction of Historic Scientific Articles](https://dl.acm.org/doi/10.1007/978-3-031-43849-3_23)".
|
15 |
|
16 |
+
Synthetic ground truth sentences have been mined from the [ar𝜒iv Bulk Downloads](https://info.arxiv.org/help/bulk_data/index.html) source documents,
|
17 |
+
and Optical Character Recognition (OCR)
|
18 |
+
sentences have been generated with the [Tesseract](https://github.com/tesseract-ocr/tesseract) OCR engine on the PDF pages generated from compiled source documents.
|
19 |
+
|
20 |
+
No page augmentation has been applied to any of the PDF documents (i.e. these are "clean" pages without warping, dust, etc.)
|
21 |
+
|
22 |
+
## Resources
|
23 |
+
|
24 |
+
### Dataset Versions
|
25 |
+
|
26 |
+
* V0 (original released with original paper) is available [here](https://zenodo.org/records/8006584)
|
27 |
+
|
28 |
+
## Citation
|
29 |
|
30 |
Please reference the following if you make use of this dataset:
|
31 |
|