OpenPecha Toolkit allows state of the art for distributed standoff annotations on moving texts
Project description
OpenPecha Toolkit
OpenPecha Toolkit allows state of the art for distributed standoff annotations on moving texts, in which Base layer can be edited without affecting annotations.
The motivation for this project it that for perfect base-text, there no big obstacles but the technical problems come in when you have to be able to edit the base-text, which can be correcting or updating the base-text. So the existing solution like using character coordinates won’t work. So we purposed the CCTV (Character Coordinate Translation Vector) to track the annotations from source base-text to edited base-text without worrying about the annotations at all. Then user can export the edited based text with updated annotations in various docuemnt format like .md
, .epub
, .pdf
, etc. But currently it supports only markdown file.
For NLP this toolkit will provide a way to have annoated corpra with minimal errors and extract a particular type of annotation or collection of different type of annotations. NLP researchers can then use these corpus to build language model, annotations to build NER model, entity linking, ect.
Prerequisite
- Python3, you can download from here
Installation
Usage
First, we need to download all the pecha from OpenPecha.
$ openpecha download --help
Usage: openpecha download [OPTIONS]
Command to download poti. You need to give a work-id of a poti to download it.
Options:
-n, --number WORK_ID Work-id of the poti, for single poti download
--help Show this message
Automatic updating annotations from source base-text (original) and destination base-text (edited)
$ openpecha update --help
Usage: openpecha update [OPTIONS] WORK_ID
Command to update the base text with your edits.
Options:
--help Show this message and exit.
Exporting and Extracting layer
$ openpecha layer --help
Usage: openpecha layer [OPTIONS] WORK_ID OUT
Command to apply a single layer, multiple layers or all available layers
(by default) and then export to markdown.
Args:
- WORK_ID is the work-id of the poti, from which given layer will be
applied
- OUT is the filename to the write the result. Currently support only
Markdown file.
Options:
-n, --name [title|tsawa|yigchung|quotes|sapche]
name of a layer to be applied
-l, --list TEXT list of name of layers to applied,
name of layers should be comma separated
--help Show this message and exit.
Developer Installation.
$ git clone https://github.com/OpenPoti/openpecha-toolkit.git
$ cd openpecha-toolkit
$ pip install -r requirements.txt
$ pip install -e .
Testing
$ pytest tests
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for openpecha-0.7.66-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | eff4cfc25f74d8a6de9fa669948ec2a0b4a018193311a3b0e06aba8a8f60e35c |
|
MD5 | fb22d6a7052c0acc7aab756191400c1b |
|
BLAKE2b-256 | 299c7b1accf89813a74a1fb98ef1892cc289895c184e1a37e492df99ed6595fd |