qurator-spk
/
neat
mirror of https://github.com/qurator-spk/neat.git
1
0
Fork 0
Code Issues Releases Wiki Activity
You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
master
Branches Tags
${ item.name }
Create tag ${ searchTerm }
Create branch ${ searchTerm }
from '07cae62901'
${ noResults }
neat/docs/Provenance.md

46 lines
3.3 KiB
Markdown
Raw Normal View History Unescape Escape

Update Provenance.md
5 years ago
# Provenance
rename Preprocessing.md -> Provenance.md
5 years ago
Update Provenance.md
5 years ago
The processing pipeline that is developed at the
rename Preprocessing.md -> Provenance.md
5 years ago
[Berlin State Library](http://staatsbibliothek-berlin.de/)
comprises the following steps:
### Layout Analysis & Textline Extraction
Layout Analysis & Textline Extraction @[sbb_textline_detector](https://github.com/qurator-spk/sbb_textline_detector)
``INPUT ``: image file
``OUTPUT``: [PAGE-XML](https://github.com/PRImA-Research-Lab/PAGE-XML) file with bounding boxes for regions and text lines
### OCR & Word Segmentation
OCR is based on [OCR-D](https://github.com/OCR-D)'s [ocrd_tesserocr](https://github.com/OCR-D/ocrd_tesserocr) which requires [Tesseract](https://github.com/tesseract-ocr/tesseract) **>= 4.1.0**. The [GT4HistOCR_2000000](https://ub-backup.bib.uni-mannheim.de/~stweil/ocrd-train/data/GT4HistOCR_2000000.traineddata) model, which is [trained](https://github.com/tesseract-ocr/tesstrain/wiki/GT4HistOCR) on the [GT4HistOCR](https://zenodo.org/record/1344132) corpus, is used. Further details are available in the [paper](https://arxiv.org/abs/1809.05501).
``INPUT ``: [PAGE-XML](https://github.com/PRImA-Research-Lab/PAGE-XML) file with bounding boxes for regions and text lines
``OUTPUT``: [PAGE-XML](https://github.com/PRImA-Research-Lab/PAGE-XML) file with bounding boxes for words and the contained text
### TSV Transformation
Update Provenance.md
5 years ago
A simple [Python tool](https://github.com/qurator-spk/page2tsv) is used for the transformation of the OCR results in [PAGE-XML](https://github.com/PRImA-Research-Lab/PAGE-XML) to [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format).
rename Preprocessing.md -> Provenance.md
5 years ago
``INPUT ``: [PAGE-XML](https://github.com/PRImA-Research-Lab/PAGE-XML) file with bounding boxes for words and the contained text
``OUTPUT``: [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format) file in the desired format for [neath](https://github.com/qurator-spk/neath)
### Tokenization
For tokenization, [SoMaJo](https://github.com/tsproisl/SoMaJo) is used.
``INPUT ``: [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format) file in the desired format for [neath](https://github.com/qurator-spk/neath)
``OUTPUT``: [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format) file with tokenization
### Named Entity Recognition
For Named Entity Recognition, a [BERT-Base](https://github.com/google-research/bert) model was trained for noisy OCR texts with historical spelling variation. [sbb_ner](https://github.com/qurator-spk/sbb_ner) is using a combination of unsupervised training on a large (~2.3m pages) [corpus of German OCR](https://zenodo.org/record/3257041) in combination with supervised training on a small (47k tokens) [annotated corpus](https://github.com/EuropeanaNewspapers/ner-corpora/tree/master/enp_DE.sbb.bio). Further details are available in the [paper](https://corpora.linguistik.uni-erlangen.de/data/konvens/proceedings/papers/KONVENS2019_paper_4.pdf).
``INPUT ``: [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format) file obtained after [Tokenization](https://github.com/qurator-spk/neath/blob/master/docs/Preprocessing.md#tokenization) and postprocessing
``OUTPUT``: [TSV](https://github.com/qurator-spk/neath/blob/master/docs/User_Guide.md#data-format) file with automatically recognized named entities added