language:
- it
license: cc-by-nc-sa-3.0
pretty_name: paisà
dataset_info:
features:
- name: text_id
dtype: int64
- name: url
dtype: string
- name: text_content
dtype: string
splits:
- name: train
num_bytes: 1166723608.8
num_examples: 308484
- name: test
num_bytes: 291680902.2
num_examples: 77121
download_size: 926608734
dataset_size: 1458404511
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
Link to the project: https://www.corpusitaliano.it/it/index.html
The PAISÀ corpus is an extensive collection of authentic Italian language texts taken from the Internet. The compilation of contemporary texts was created within the framework of the PAISÀ project (Platform for Learning Italian through Annotated Corpora) with the aim of providing authentic and freely available material for learning Italian.
A unique resource for the Italian language has emerged, as it combines several features:
- Texts taken from the web (collected in September/October 2010)
- Composed entirely of texts available and reusable for free (protected by Creative Commons licenses attribution - share alike and attribution - share alike - non-commercial)
- Size: approximately 250 million tokens
- Fully annotated in CoNLL format (lemmatization and annotation of parts of speech (or also here) and syntactic dependencies)
- Preprocessed automatically and partially corrected manually (various steps performed: retrieval of URL addresses, preprocessing of collected texts, and correction of annotations for the calibration of annotation tools)
Although primarily created for learning, the PAISÀ corpus also serves as a valuable resource for various linguistic research activities.
The website aims to provide an interface for learners to access the corpus online. There are various access modes designed for different user groups, from pre-set searches to the possibility of defining search criteria in a completely flexible manner by creating complex queries. It is currently a work in continuous update and progress.
In these pages, you will also find information about the PAISÀ project, details on how the text collection was formed, and downloadable versions of the entire corpus.
The PAISÀ project is funded by the Ministry of Education, Universities, and Research (MIUR) through the Basic Research Investment Fund (FIRB).