---
title: "LocRegen: Cost-Efficient Redundancy Removal in — Zhang 2026"
description: "Online store product titles often repeat the same content words."
authors: ["Bryan Zhang", "Stephan Walter", "Luca Lomanto", "Merve Arinik"]
year: 2026
venue: "Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1)"
source: "https://aclanthology.org/2026.eamt-1.33/"
licence: "CC BY-ND 4.0 (EAMT)"
licence_url: "https://creativecommons.org/licenses/by-nd/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/zhang-2026-locregen-cost-efficient-redundancy-removal/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/zhang-2026-locregen-cost-efficient-redundancy-removal/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2026](https://md.research.elamur.ai/free/en/papers/2026/page/6/) / LocRegen: Cost-Efficient Redundancy Removal in Multilingual E-commerce Titles with Small Language Models

# LocRegen: Cost-Efficient Redundancy Removal in Multilingual E-commerce Titles with Small Language Models

**Authors:** Bryan Zhang, Stephan Walter, Luca Lomanto, Merve Arinik · **Year:** 2026 · **Venue:** Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1) · **Licence:** [CC BY-ND 4.0 (EAMT)](https://creativecommons.org/licenses/by-nd/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/2026.eamt-1.33/) · [PDF](https://aclanthology.org/2026.eamt-1.33.pdf)

## Summary
Online store product titles often repeat the same content words. Sellers add keywords for search optimization, and machine translation also introduces repetition. The authors first examine why localized titles become redundant and identify three linguistic causes: compounds being split during translation (e.g., German, Swedish), several source words mapping to a single target word, and loss of morphological distinctions such as grammatical gender.

They then present LocRegen, a system that lets a small language model rewrite titles so that no content word occurs more than twice, while keeping the attributes required by product-type title templates. It has three parts:
- a multilingual Repeated Word Detection (RWD) module that reports remaining repetitions,
- a Dynamic Feedback-aware Regeneration (DFR) loop that retries with progressively stricter prompts,
- "smart cues," which tag the first two occurrences of a repeated word so the model keeps them.

In a manual evaluation across five languages, a quantized 7B model running inside LocRegen beat a 47B mixture-of-experts model on both remaining redundancy and errors. This shows that industrial-scale title cleanup can run on inexpensive hardware.

## Key points
- A repeated word is a content word (function words excluded) appearing more than twice. Localization made 19%–50% of non-redundant source titles redundant, depending on the language pair: German–Spanish was highest at 50% and English–French lowest at 19%. Each pair was measured on a sample of over 10K titles.
- LocRegen (7B) compared with the 47B MoE baseline:
  - redundant title rate: 2.4% vs. 3.5%
  - overall error rate, including omission of key attributes: 3.8% vs. 8.4%
  - title length reduction with LocRegen: 27.5–32.1%
- DFR repeats regeneration until no repetitions remain or an iteration limit is reached. It moves through three prompt templates (vanilla, mid-aggressive, aggressive). Smart cues are applied in rounds 1–5, and product-type title templates in rounds 1–3.
- The system is designed to run on a single NVIDIA T4 GPU (16 GiB). The experiments themselves used a machine with 8 NVIDIA A10G GPUs.

*Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.*

Fields: [Linguistics & languages](https://md.research.elamur.ai/free/en/shelves/linguistics/) · [Machine translation](https://md.research.elamur.ai/free/en/shelves/machine-translation/)

## Cite

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{zhang2026locregen,
  title = {{LocRegen: Cost-Efficient Redundancy Removal in Multilingual E-commerce Titles with Small Language Models}},
  author = {Zhang, Bryan and Walter, Stephan and Lomanto, Luca and Arinik, Merve},
  booktitle = {Proceedings of the 26th Annual Conference of the European Association for Machine Translation (Volume 1)},
  year = {2026},
  url = {https://aclanthology.org/2026.eamt-1.33/}
}
```

- [← previous LLMs Out-of-the-Box Do Not Generate Context-Appropriate Word Order in Russian](https://md.research.elamur.ai/free/en/papers/shabaeva-2026-llms-out-box-do-not/)
- [next → Logic-Level Evaluation of Logical Table-to-Text Generation](https://md.research.elamur.ai/free/en/papers/trigg-2026-logic-level-evaluation-logical-table/)
- [↑ back to papers of 2026](https://md.research.elamur.ai/free/en/papers/2026/page/6/)

## References

23 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Bi, Tianchi, Liang Yao, Baosong Yang, Haibo Zhang, Weihua Luo, and Boxing Chen. 2020. Constraint translation candidates: A bridge between neural query translation and cross-lingual information retrieval. [find](https://scholar.google.com/scholar?q=Bi%2C+Tianchi%2C+Liang+Yao%2C+Baosong+Yang%2C+Haibo+Zhang%2C+Weihua+Luo%2C+and+Boxing+Chen%2E+2020%2E+Constraint+translation+candidates%3A+A+bridge+between+neural+query+translation+and+cross%2Dlingual+information+retriev)
2. Fetahu, Besnik, Zhiyu Chen, Oleg Rokhlenko, and Shervin Malmasi. 2023. InstructPTS: Instruction-tuning LLMs for product title summarization. In Wang, Mingxuan and Imed Zitouni, editors, Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing: Industry Track, pages 663–… [find](https://scholar.google.com/scholar?q=Fetahu%2C+Besnik%2C+Zhiyu+Chen%2C+Oleg+Rokhlenko%2C+and+Shervin+Malmasi%2E+2023%2E+InstructPTS%3A+Instruction%2Dtuning+LLMs+for+product+title+summarization%2E+In+Wang%2C+Mingxuan+and+Imed+Zitouni%2C+editors%2C+Proceedings+of)
3. Guan, Xinyi, Shun Long, Weiheng Zhu, Silei Cao, and Fangting Liao. 2022. Mask-based text scoring for product title summarization. In 2022 8th International Conference on Systems and Informatics (ICSAI), pages 1–6. [find](https://scholar.google.com/scholar?q=Guan%2C+Xinyi%2C+Shun+Long%2C+Weiheng+Zhu%2C+Silei+Cao%2C+and+Fangting+Liao%2E+2022%2E+Mask%2Dbased+text+scoring+for+product+title+summarization%2E+In+2022+8th+International+Conference+on+Systems+and+Informatics+%28ICSAI)
4. Guha, Jyoti and Carmen Heger. 2014. Machine translation for global e-commerce on ebay. In Proceedings of the AMTA, volume 2, pages 31–37. [find](https://scholar.google.com/scholar?q=Guha%2C+Jyoti+and+Carmen+Heger%2E+2014%2E+Machine+translation+for+global+e%2Dcommerce+on+ebay%2E+In+Proceedings+of+the+AMTA%2C+volume+2%2C+pages+31%E2%80%9337%2E)
5. Jiang, Zhuolin, Amro El-Jaroudi, William Hartmann, Damianos Karakos, and Lingjun Zhao. 2020. Cross-lingual information retrieval with BERT. In Proceedings of the workshop on Cross-Language Search and Summarization of Text and Speech (CLSSTS2020), pages 26–31, Marseille, France, May. European Languag… [find](https://scholar.google.com/scholar?q=Jiang%2C+Zhuolin%2C+Amro+El%2DJaroudi%2C+William+Hartmann%2C+Damianos+Karakos%2C+and+Lingjun+Zhao%2E+2020%2E+Cross%2Dlingual+information+retrieval+with+BERT%2E+In+Proceedings+of+the+workshop+on+Cross%2DLanguage+Search+and+)
6. Jiang, Albert Q., Alexandre Sablayrolles, Antoine Roux, Arthur Mensch, Blanche Savary, Chris Bamford, Devendra Singh Chaplot, Diego de las Casas, Emma Bou Hanna, Florian Bressand, Gianna Lengyel, Guillaume Bour, Guillaume Lample, Lélio Renard Lavaud, Lucile Saulnier, Marie-Anne Lachaux, Pierre Stoc… [find](https://scholar.google.com/scholar?q=Jiang%2C+Albert+Q%2E%2C+Alexandre+Sablayrolles%2C+Antoine+Roux%2C+Arthur+Mensch%2C+Blanche+Savary%2C+Chris+Bamford%2C+Devendra+Singh+Chaplot%2C+Diego+de+las+Casas%2C+Emma+Bou+Hanna%2C+Florian+Bressand%2C+Gianna+Lengyel%2C+Guil)
7. Lakew, Surafel Melaku, Mattia Di Gangi, and Marcello Federico. 2019. Controlling the output length of neural machine translation. In Niehues, Jan, Rolando Cattoni, Sebastian Stüker, Matteo Negri, Marco Turchi, Thanh-Le Ha, Elizabeth Salesky, Ramon Sanabria, Loic Barrault, Lucia Specia, and Marcello… [find](https://scholar.google.com/scholar?q=Lakew%2C+Surafel+Melaku%2C+Mattia+Di+Gangi%2C+and+Marcello+Federico%2E+2019%2E+Controlling+the+output+length+of+neural+machine+translation%2E+In+Niehues%2C+Jan%2C+Rolando+Cattoni%2C+Sebastian+Stu%CC%88ker%2C+Matteo+Negri%2C+Mar)
8. Lowndes, Mike and Aditya Vasudevan. 2021. Market guide for digital commerce search. [find](https://scholar.google.com/scholar?q=Lowndes%2C+Mike+and+Aditya+Vasudevan%2E+2021%2E+Market+guide+for+digital+commerce+search%2E)
9. Nida, Eugene Albert and Charles Russell Taber. 1974. The theory and practice of translation, volume 8. Brill Archive. [find](https://scholar.google.com/scholar?q=Nida%2C+Eugene+Albert+and+Charles+Russell+Taber%2E+1974%2E+The+theory+and+practice+of+translation%2C+volume+8%2E+Brill+Archive%2E)
10. Nie, Jian-Yun. 2010. Cross-language information retrieval. Synthesis Lectures on Human Language Technologies, 3(1):1–125. [find](https://scholar.google.com/scholar?q=Nie%2C+Jian%2DYun%2E+2010%2E+Cross%2Dlanguage+information+retrieval%2E+Synthesis+Lectures+on+Human+Language+Technologies%2C+3%281%29%3A1%E2%80%93125%2E)
11. Rozen, Ohad, David Carmel, Avihai Mejer, Vitaly Mirkis, and Yftah Ziser. 2021. Answering product-questions by utilizing questions from other contextually similar products. In Toutanova, Kristina, Anna Rumshisky, Luke Zettlemoyer, Dilek Hakkani-Tur, Iz Beltagy, Steven Bethard, Ryan Cotterell, Tanmoy… [find](https://scholar.google.com/scholar?q=Rozen%2C+Ohad%2C+David+Carmel%2C+Avihai+Mejer%2C+Vitaly+Mirkis%2C+and+Yftah+Ziser%2E+2021%2E+Answering+product%2Dquestions+by+utilizing+questions+from+other+contextually+similar+products%2E+In+Toutanova%2C+Kristina%2C+Anna)
12. Rücklé, Andreas, Krishnkant Swarnkar, and Iryna Gurevych. 2019. Improved cross-lingual question retrieval for community question answering. In The World Wide Web Conference, WWW ’19, page 3179–3186, New York, NY, USA. Association for Computing Machinery. [find](https://scholar.google.com/scholar?q=Ru%CC%88ckle%CC%81%2C+Andreas%2C+Krishnkant+Swarnkar%2C+and+Iryna+Gurevych%2E+2019%2E+Improved+cross%2Dlingual+question+retrieval+for+community+question+answering%2E+In+The+World+Wide+Web+Conference%2C+WWW+%E2%80%9919%2C+page+3179%E2%80%933186%2C)
13. Saleh, Shadi and Pavel Pecina. 2020. Document translation vs. query translation for cross-lingual information retrieval in the medical domain. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6849–6860, Online, July. Association for Computational Ling… [find](https://scholar.google.com/scholar?q=Saleh%2C+Shadi+and+Pavel+Pecina%2E+2020%2E+Document+translation+vs%2E+query+translation+for+cross%2Dlingual+information+retrieval+in+the+medical+domain%2E+In+Proceedings+of+the+58th+Annual+Meeting+of+the+Associat)
14. Samar, Thaer, Myriam C. Traub, Jacco Ossenbruggen, Lynda Hardman, and Arjen P. Vries. 2018. Quantifying retrieval bias in web archive search. Int. J. Digit. Libr., 19(1):57–75, mar. [find](https://scholar.google.com/scholar?q=Samar%2C+Thaer%2C+Myriam+C%2E+Traub%2C+Jacco+Ossenbruggen%2C+Lynda+Hardman%2C+and+Arjen+P%2E+Vries%2E+2018%2E+Quantifying+retrieval+bias+in+web+archive+search%2E+Int%2E+J%2E+Digit%2E+Libr%2E%2C+19%281%29%3A57%E2%80%9375%2C+mar%2E)
15. Sun, Fei, Peng Jiang, Hanxiao Sun, Changhua Pei, Wenwu Ou, and Xiaobo Wang. 2018. Multi-source pointer network for product title summarization. In Proceedings of the 27th ACM International Conference on Information and Knowledge Management, CIKM ’18, page 7–16, New York, NY, USA. Association for Com… [find](https://scholar.google.com/scholar?q=Sun%2C+Fei%2C+Peng+Jiang%2C+Hanxiao+Sun%2C+Changhua+Pei%2C+Wenwu+Ou%2C+and+Xiaobo+Wang%2E+2018%2E+Multi%2Dsource+pointer+network+for+product+title+summarization%2E+In+Proceedings+of+the+27th+ACM+International+Conference+)
16. Team, Qwen. 2024. Qwen2.5: A party of foundation models, September. [find](https://scholar.google.com/scholar?q=Team%2C+Qwen%2E+2024%2E+Qwen2%2E5%3A+A+party+of+foundation+models%2C+September%2E)
17. Wang, Manyi, Tao Zhang, Qijin Chen, and Chengfu Huo. 2020. Selling products by machine: a user-sensitive adversarial training method for short title generation in mobile e-commerce. [find](https://scholar.google.com/scholar?q=Wang%2C+Manyi%2C+Tao+Zhang%2C+Qijin+Chen%2C+and+Chengfu+Huo%2E+2020%2E+Selling+products+by+machine%3A+a+user%2Dsensitive+adversarial+training+method+for+short+title+generation+in+mobile+e%2Dcommerce%2E)
18. Wang, Haifeng, Hua Wu, Zhongjun He, Liang Huang, and Kenneth Ward Church. 2021. Progress in machine translation. Engineering. [find](https://scholar.google.com/scholar?q=Wang%2C+Haifeng%2C+Hua+Wu%2C+Zhongjun+He%2C+Liang+Huang%2C+and+Kenneth+Ward+Church%2E+2021%2E+Progress+in+machine+translation%2E+Engineering%2E)
19. Way, Andy. 2013. Traditional and emerging use-cases for machine translation. Proceedings of Translating and the Computer, 35:12. [find](https://scholar.google.com/scholar?q=Way%2C+Andy%2E+2013%2E+Traditional+and+emerging+use%2Dcases+for+machine+translation%2E+Proceedings+of+Translating+and+the+Computer%2C+35%3A12%2E)
20. Xiao, Joan and Robert Munro. 2019. Text summarization of product titles. In eCOM@SIGIR. [find](https://scholar.google.com/scholar?q=Xiao%2C+Joan+and+Robert+Munro%2E+2019%2E+Text+summarization+of+product+titles%2E+In+eCOM%40SIGIR%2E)
21. Zhang, Xueying, Yunjiang Jiang, Yue Shang, Zhaomeng Cheng, Chi Zhang, Xiaochuan Fan, Yun Xiao, and Bo Long. 2021. Dsgpt: Domain-specific generative pre-training of transformers for text generation in e-commerce title and review summarization. In Proceedings of the 44th International ACM SIGIR Confer… [find](https://scholar.google.com/scholar?q=Zhang%2C+Xueying%2C+Yunjiang+Jiang%2C+Yue+Shang%2C+Zhaomeng+Cheng%2C+Chi+Zhang%2C+Xiaochuan+Fan%2C+Yun+Xiao%2C+and+Bo+Long%2E+2021%2E+Dsgpt%3A+Domain%2Dspecific+generative+pre%2Dtraining+of+transformers+for+text+generation+in+)
22. Zhang, Bryan, Taichi Nakatani, Daniel Vidal Hussey, Stephan Walter, and Liling Tan. 2024. Don’t just translate, summarize too: Cross-lingual product title generation in E-commerce. In Malmasi, Shervin, Besnik Fetahu, Nicola Ueffing, Oleg Rokhlenko, Eugene Agichtein, and Ido Guy, editors, Proceedings… [find](https://scholar.google.com/scholar?q=Zhang%2C+Bryan%2C+Taichi+Nakatani%2C+Daniel+Vidal+Hussey%2C+Stephan+Walter%2C+and+Liling+Tan%2E+2024%2E+Don%E2%80%99t+just+translate%2C+summarize+too%3A+Cross%2Dlingual+product+title+generation+in+E%2Dcommerce%2E+In+Malmasi%2C+Shervin)
23. Zhou, Mingyang, Runxiang Cheng, Yong Jae Lee, and Zhou Yu. 2018. A visual attention grounding neural model for multimodal machine translation. CoRR, abs/1808.08266. [find](https://scholar.google.com/scholar?q=Zhou%2C+Mingyang%2C+Runxiang+Cheng%2C+Yong+Jae+Lee%2C+and+Zhou+Yu%2E+2018%2E+A+visual+attention+grounding+neural+model+for+multimodal+machine+translation%2E+CoRR%2C+abs%2F1808%2E08266%2E)