Generate large textual corpora for almost any language by crawling the web
webcorpus is an end-to-end tool used to crawl and generate datasets from the crawled data. It can be used to generate monolingual corpora and has various processors to create labelled datasets automatically. webcorpus is particulary suited for low-resource languages which need automated methods for creating large-scale datasets.
Make sure you have java installed on your system. Next, install it using pip:
sudo pip3 install webcorpus
To build the dataset, we first need to crawl the web and then process the crawls to create the final dataset.
Step 1: Crawling Sources
To start crawling websites, you first need to start the webcorpus crawling server:
Once the server has started, you can start crawls using the following command.
webcorpus crawl --path <path> --name <name> --url <url> --log <path> [--host <ip address>]
You can see the status of the crawls anytime by executing:
webcorpus log [--host <ip address>]
The last two steps can also been remotely, which can be useful in distributed mode where you are running multiple webcorpus servers.
Step 2: Processing Corpus
webcorpus process --operation <operation code> --lang <lang code> --input <input path> --output <output path>
Currently, the following processing operations are supported:
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size webcorpus-0.2-py3-none-any.whl (55.1 kB)||File type Wheel||Python version py3||Upload date||Hashes View|
|Filename, size webcorpus-0.2.tar.gz (35.4 kB)||File type Source||Python version None||Upload date||Hashes View|