BrWaC

The Brazilian Portuguese Web as Corpus is a large corpus constructed in our lab following the Wacky framework, which was made public for research purposes.

Current version

The current corpus version, released in January 2017, is composed by 3.53 million documents, 2.68 billion tokens and 5.79 million types (TTR 0.0021), and was completely parsed with the Palavras parser. It is now openly available, both for download and for navigation in a NoSketch Engine interface, and you can request access here. If you have any problems, please contact us directly through brwacteam at gmail.

Classification according to readability levels

During our studies, the whole corpus was annotated with 134 different textual attributes and classified according to readability levels by four different learning models with different characteristics. For more information about these experiments, please check our publications or contact us.

Publications

• Jorge Alberto Wagner Filho, Rodrigo Wilkens, Marco Idiart and Aline Villavicencio. The brWaC Corpus: A New Open Resource for Brazilian Portuguese. In 11th Language Resources and Evaluation Conference (LREC). (to appear). 🇬🇧 (Main reference)

• Jorge Alberto Wagner Filho, Rodrigo Wilkens and Aline Villavicencio. Automatic Construction of Large Readability Corpora. In 1st Workshop on Computational Linguistics for Linguistic Complexity (CL4LC). (PDF). 🇬🇧

• Jorge A. Wagner Filho, Rodrigo Wilkens, Leonardo Zilio, Marco Idiart, and Aline Villavicencio. Crawling by readability level. In 12th International Conference on Computational Processing of the Portuguese Language (pp. 306-318). Springer International Publishing. (PDF). 🇬🇧

• Jorge Alberto Wagner Filho. Coleta automática de corpora Web classificados por grau de legibilidade para o português. BSc. Final Project, UFRGS Digital Repository. (Lume). 🇧🇷