Concept-aware contrastive representation learning for cross-translation semantic consistency modeling of the Analects
Tang He, Peng Dai, Rong Wang, Yanyang Zhao, Chen Zhao · 2026
Read Concept-aware contrastive representation learning for cross-translation semantic consistency modeling of the Analects by Tang He, Peng Dai, Rong Wang, Yanyang Zhao, Chen Zhao (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0358227.txt.
Description
Tang He and coauthors published a 2026 study in PLOS ONE that brings natural language processing to a classic of Chinese philosophy, the Analects of Confucius. Because its many English translations differ in wording and in how they handle key concepts, He's team built a computer framework for comparing them. It lines up five English versions sentence by sentence and measures how closely they agree in meaning.
Their method cleans up concept terms and proper names first, then trains a shared sentence encoder with contrastive learning so matched sentences sit close together in the model. Tested on finding matching sentences across versions and on flagging pairs that disagree, the full model beat the alternatives they tried. Higher model scores also lined up with pairs that human judges had rated as highly consistent. They offer the tool as an aid for scholars comparing translations, not an automatic judge of quality.
Catalog record
| CREATOR | Tang He, Peng Dai, Rong Wang, Yanyang Zhao, Chen Zhao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63351 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |