In the recent years, Transformer-based models have lead to significant advances in language modelling for natural language processing. However, they require a vast amount of data to be (pre-)trained and there is a lack of corpora in languages other than English. Recently, several initiatives have presented multilingual datasets obtained from automatic web crawling. However, the results in Spanish present important shortcomings, as they are either too small in comparison with other languages, or present a low quality derived from sub-optimal cleaning and deduplication. In this paper, we introduce esCorpius, a Spanish crawling corpus obtained from near 1 Pb of Common Crawl data. It is the most extensive corpus in Spanish with this level of quality in the extraction, purification and deduplication of web textual content. Our data curation process involves a novel highly parallel cleaning pipeline and encompasses a series of deduplication mechanisms that together ensure the integrity of both document and paragraph boundaries. Additionally, we maintain both the source web page URL and the WARC shard origin URL in order to complain with EU regulations. esCorpius has been released under CC BY-NC-ND 4.0 license.
This package comprises eight models of Czech word embeddings trained by applying word2vec (Mikolov et al. 2013) to the currently most extensive corpus of Czech, namely SYN v9 (Křen et al. 2022). The minimum frequency threshold for including a word in the model was 10 occurrences in the corpus. The original lemmatisation and tagging included in the corpus were used for disambiguation. In the case of word embeddings of word forms, units comprise word forms and their tag from a positional tagset (cf. https://wiki.korpus.cz/doku.php/en:pojmy:tag) separated by '>', e.g., kočka>NNFS1-----A----.
The published package provides models trained on both tokens and lemmas. In addition, the models combine training algorithms (CBOW and Skipgram) and dimensions of the resulting vectors (100 or 500), while the training window and negative sampling remained the same during the training. The package also includes files with frequencies of word forms (vocab-frequencies.forms) and lemmas (vocab-frequencies.lemmas).