PaddyWaC: A Minimally-Supervised Web-Corpus of Hiberno-English
Bibliographic record
Abstract
Small, manually assembled corpora may be available for less dominant languages and dialects, but producing web-scale resources remains a challenge. Even when considerable quantities of text are present on the web, finding this text, and distinguishing it from related languages in the same region can be difficult. For example less dominant variants of English (e.g. New Zealander, Singaporean, Canadian, Irish, South African) may be found under their respective national domains, but will be partially mixed with Englishes of the British and US varieties, perhaps through syndication of journalism, or the local reuse of text by multinational companies. Less formal dialectal usage may be scattered more widely over the internet through mechanisms such as wiki or blog authoring. Here we automatically construct a corpus of Hiberno-English (English as spoken in Ireland) using a variety of methods: filtering by national domain, filtering by orthographic conventions, and bootstrapping from a set of Ireland-specific terms (slang, place names, organisations). We evaluate the national specificity of the resulting corpora by measuring the incidence of topical terms, and several grammatical constructions that are particular to Hiberno-English. The results show that domain filtering is very effective for isolating text that is topic-specific, and orthographic classification can exclude some non-Irish texts, but that selected seeds are necessary to extract considerable quantities of more informal, dialectal text.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.006 | 0.004 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.015 | 0.012 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".