ucrelnlp/Multilingual-USAS-Labelled-Silver-Wikipedia

Multilingual USAS Silver Labelled Wikipedia Articles

Silver-labelled Wikipedia article text for training USAS semantic taggers and Multi-Word Expression (MWE) identifiers, covering 8 Wikipedia language sites. The source text comes from the HuggingFace HuggingFaceFW/finewiki dataset, restricted to articles rated Good (GA) or Featured (FA) — using the article ID list from ucrelnlp/wikipedia-ga-fa-ids — and then sentence split and automatically tagged with USAS semantic tags and MWEs using PyMUSAS rule-based taggers. For more information on how the dataset was generated, including the full filtering/processing pipeline, see https://github.com/UCREL/wikipedia-USAS-processing.

• Curated by: University Centre for Computer Corpus Research on Language (UCREL) group at Lancaster University
• Multi-lingual
• Repository: https://github.com/UCREL/wikipedia-USAS-processing

Uses

It can be used to train USAS semantic taggers and MWE identifiers.

Filtering and Processing

Each Wikipedia article goes through the following pipeline before being included in this dataset:

• The article ID and title must match an article rated as Good (GA) or Featured (FA) (taken from ucrelnlp/wikipedia-ga-fa-ids).
• Articles that are part of a manually curated test set (by URL) are excluded.
• Wikipedia family-tree tables, mathematical equations, and other tables are removed from the article text.
• Markdown formatting (e.g. headers like # but not the header text) is stripped from the article text.
• Articles with fewer than 50 tokens, based on a language-specific tokenizer, are removed.
• Exact and then MinHash de-duplication is applied.
• Remaining articles are sentence split using language-specific spaCy sentence splitters.
• Each sentence is tagged with USAS semantic tags and, where the tagger supports it, MWEs, using PyMUSAS Rule Based language-specific taggers.

Dataset Structure

Each row is a single article, unique per id/page_id within a language config. The data is split per language into train and validation subsets (see below).

• text - the processed article text.
• id - unique identifier for the article, e.g. dawiki/1171348.
• page_id - the Wikipedia page ID, e.g. 1171348.
• title - the article title.
• url - the article URL, e.g. https://da.wikipedia.org/wiki/El_Salvador_ved_sommer-OL_2024.
• version (int|string) - revision/version identifier of the page, as provided by HuggingFaceFW/finewiki, e.g. 1167219203.
• start_end_sentence_character_indexes - list of [start, end] character offsets for each sentence, e.g. [[0, 10], [11, 15]] — the first sentence is text[0:10].
• tokens - list of a list of tokens, where the inner list is the tokens for a given sentence, e.g. tokens[0] contains all tokens in the first sentence.
• tags - list of a list of a list of USAS tags predicted by the PyMUSAS rule-based tagger. The innermost list holds the most likely USAS tag(s) for a given token, e.g. tags[0][0] is the most likely USAS tag(s) for the first token of the first sentence — usually a single tag, but more than one when the token's meaning is a combination of the predicted tags. Some tokens have no USAS tags, as the rule-based tagger cannot make a prediction for every token.
• other_tags - list of a list of a list of USAS tags, same shape as tags, containing every other valid USAS tag for a token that was not among its most likely tag(s), e.g. other_tags[0][0]. Most tokens have no other tags, in which case the inner list is empty.
• mwes - list of a list of MWE labels predicted by the PyMUSAS rule-based tagger, relating to the most likely USAS tags. Labels denote, at the sentence level, which tokens form an MWE, e.g. if mwes[0][0] and mwes[0][1] both contain 1, the first and second tokens of the first sentence are part of the same MWE. MWEs can be discontinuous but should not overlap. MWE label indexes start at 1 and reset per sentence — MWEs cannot span sentence boundaries.

The data is stored as zstd-compressed Parquet files.

Example of a record (shown as JSON for readability):

{
  "text": "El Salvador deltog i sommer-OL 2024 i Paris.",
  "id": "dawiki/1171348",
  "page_id": 1171348,
  "title": "El Salvador ved sommer-OL 2024",
  "url": "https://da.wikipedia.org/wiki/El_Salvador_ved_sommer-OL_2024",
  "version": 1167219203,
  "start_end_sentence_character_indexes": [[0, 45]],
  "tokens": [["El", "Salvador", "deltog", "i", "sommer-OL", "2024", "i", "Paris", "."]],
  "tags": [[["Z2"], ["Z2"], ["M6"], ["Z5"], ["K5.1"], ["T1.3"], ["Z5"], ["Z2"], ["Z9"]]],
  "other_tags": [[[], [], [], [], [], [], [], [], []]],
  "mwes": [[1, 1, 0, 0, 0, 0, 0, 0, 0]]
}

Train/validation split

Each language's documents are split into train and validation subsets, written to separate train/validation subfolders. The validation split is capped at whichever is reached first: a percentage of the language's documents, or a fixed maximum number of documents — this keeps under-resourced languages (some have as few as ~200 articles) at a sensible percentage-based split, while bounding well-resourced languages' validation set to a sane absolute size. The split assignment is deterministic (hashed from each document's page ID), so re-running the pipeline reproduces the same split.

Dataset Statistics

The table below shows per language the number of entries/articles that are either Good or Featured (Total), Good, or Featured:

License

This dataset contains text from Wikipedia, licensed under Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0) and also available under GFDL. See Wikipedia’s licensing and Terms of Use: https://dumps.wikimedia.org/legal.html

We release this data under the same license; Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0) and also available under GFDL.

Dataset Card Authors

• UCREL (ucrel@lancaster.ac.uk)
• Andrew Moore / apmoore1 (a.p.moore@lancaster.ac.uk / andrew.p.moore94@gmail.com)
• Paul Rayson (p.rayson@lancaster.ac.uk)

Dataset Card Contact

• UCREL (ucrel@lancaster.ac.uk)
• Andrew Moore / apmoore1 (a.p.moore@lancaster.ac.uk / andrew.p.moore94@gmail.com)
• Paul Rayson (p.rayson@lancaster.ac.uk)