{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/guided-alignment-training-for-topic-aware","title":"Guided Alignment Training for Topic-Aware Neural Machine Translation","arxiv_id":"1607.01628","date":"2016-07-06","proceeding":"AMTA 2016 10","authors":["Wenhu Chen","Evgeny Matusov","Shahram Khadivi","Jan-Thorsten Peter"],"abstract":"In this paper, we propose an effective way for biasing the attention\nmechanism of a sequence-to-sequence neural machine translation (NMT) model\ntowards the well-studied statistical word alignment models. We show that our\nnovel guided alignment training approach improves translation quality on\nreal-life e-commerce texts consisting of product titles and descriptions,\novercoming the problems posed by many unknown words and a large type/token\nratio. We also show that meta-data associated with input texts such as topic or\ncategory information can significantly improve translation quality when used as\nan additional signal to the decoder part of the network. With both novel\nfeatures, the BLEU score of the NMT system on a product title set improves from\n18.6 to 21.3%. Even larger MT quality gains are obtained through domain\nadaptation of a general domain NMT system to e-commerce data. The developed NMT\nsystem also performs well on the IWSLT speech translation task, where an\nensemble of four variant systems outperforms the phrase-based baseline by 2.1%\nBLEU absolute.","url_abs":"http://arxiv.org/abs/1607.01628v1","url_pdf":"http://arxiv.org/pdf/1607.01628v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"guided-alignment-training-for-topic-aware","repo_url":"https://github.com/OpenNMT/OpenNMT-tf","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"domain-adaptation","task_name":"Domain Adaptation"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"nmt","task_name":"NMT"},{"task_slug":"translation","task_name":"Translation"},{"task_slug":"word-alignment","task_name":"Word Alignment"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1607.01628","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}