WeBiText: building large heterogeneous translation memories from parallel Web content
Bibliographic record
Abstract
This paper investigates the extent to which a useful general purpose Translation Memory (TM) can be built based on very large amounts of heterogeneous parallel texts mined from the Web. In particular, we evaluate whether such a TM could add value over TMs built from other large, publicly available parallel corpora, such as the Canadian Hansard. In the case of Canadian translators working with English and French, we show that the answer to both questions is a resounding yes. Using field data collected through contextualized observation and interviews with translators at their workplace, we show how this concept is well grounded in existing workpractices of translators, especially Canadian ones. We also show that a TM based on 10 million pairs of pages from Government of Canada Web sites is able to cover 90% of the translation problems observed in our interview subjects. This turns out to be significantly better than coverage of a general purpose TM built from a smaller corpus, namely, the Canadian Hansard. The difference is most notable for the harder problems, such as specialized terminology. We also evaluate the approach on Web parallel corpora for other languages (European Commission Web sites, and 5000 Inuktitut-English pages harvested from the Nunavut domain), and find the approach to not be as advantageous there. We conclude that, while the concept of building TMs from Web corpora holds great promise, more research may be needed to make it work for language pairs other than English-French.
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 | 0.012 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.004 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.006 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.007 | 0.006 |
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".