[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-tackle-a-tag-problem-in-ai-translation":10,"sections":40},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},7854,"researchers-tackle-a-tag-problem-in-ai-translation","Researchers Tackle a Tag Problem in AI Translation","A new unpublished preprint proposes a three-part fix for AI translation systems that mangle format tags while chasing fluent prose.","A new research paper attacks a very specific headache in AI translation: keeping formatting tags intact without turning the output into stilted, tag-cluttered prose.\n\nThe work, described in a preprint posted to arXiv on September 25, 2026 (arXiv:2609.29131) and not yet peer-reviewed, targets what the authors call tag-aware structured text translation - the challenge of translating text full of HTML, markdown, or other format tags while keeping both the tags and the sentence fluent. The researchers found that existing methods for generating training data optimize for either tag diversity or natural-sounding translations, but not both. Their fix, a hybrid data-synthesis method called Hy-LST, combines two tag-generation techniques to produce training examples with both varied tags and natural phrasing. They then split the translation task into four sub-tasks of rising difficulty for fine-tuning, and layered on three separate reward signals - covering fluency, tag accuracy, and translation quality within tagged spans - optimized together using a technique called group relative policy optimization. Tested across six language pairs, including English-to-Chinese, English-to-Japanese, and German-to-French, the combined system beat existing approaches, with the authors reporting a measurable gain from each of the three components on its own.\n\nThis is the kind of unglamorous problem that matters more than it sounds: any tool that runs product listings, help docs, or web pages through machine translation runs into it, since one dropped or misplaced tag can break a page's layout or strip out a hyperlink. Most public LLM-based translators still treat this as a training afterthought, so a systematic fix spanning data, fine-tuning, and reward design is a genuinely useful contribution rather than an incremental tweak.\n\nIt's still a single preprint with no independent replication or peer review, and all six tested language pairs are high-resource ones - so don't expect this to fix messier, lower-resource translation tasks anytime soon.","[\"ai\",\"machine-translation\",\"research\",\"nlp\"]","2026-09-25T04:00:00.000Z","2026-09-26T03:00:45.959Z","2026-09-26T03:00:51.029Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Add basic sourcing: cite the arXiv preprint by ID\u002Flink (arXiv:2609.29131), note it's an unpublished\u002Fnon-peer-reviewed preprint, and include the submission date, since currently the piece never attributes the research to a paper, institution, or authors.","resolved","ai",[30,32,33,34],"machine-translation","research","nlp",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.29131",0,{"sections":41},[42,46,51,56,61,66,71,76,81,86,91,96,101,106],{"name":43,"slug":30,"count":44,"latest_published_at":45},"AI",4557,"2026-09-25T17:16:30.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Security","security",741,"2026-09-25T15:52:13.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Policy","policy",390,"2026-09-25T16:24:59.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Deals","deals",256,"2026-09-25T17:00:53.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Hardware","hardware",185,"2026-09-25T15:00:22.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Science","science",140,"2026-09-25T11:55:23.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Consumer Tech","consumer-tech",132,"2026-09-25T15:30:00.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Software","software",88,"2026-09-24T23:06:55.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Dev Tools","dev-tools",82,"2026-09-25T09:59:40.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"Startups","startups",76,"2026-09-25T18:33:59.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Gaming","gaming",48,"2026-09-25T18:35:21.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"General","general",46,"2026-09-25T02:12:57.000Z",{"name":102,"slug":103,"count":104,"latest_published_at":105},"Reviews","reviews",30,"2026-09-24T20:07:31.000Z",{"name":107,"slug":108,"count":109,"latest_published_at":110},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]