@inproceedings{kulkarni-etal-2023-domain,
title = "Domain-specific transformer models for query translation",
author = "Kulkarni, Mandar and
Garera, Nikesh and
Trivedi, Anusua",
editor = "Sitaram, Sunayana and
Beigman Klebanov, Beata and
Williams, Jason D",
booktitle = "Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track)",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.acl-industry.10",
doi = "10.18653/v1/2023.acl-industry.10",
pages = "89--95",
abstract = "Due to the democratization of e-commerce, many product companies are listing their goods for online shopping. For periodic buying within a domain such as Grocery, consumers are generally inclined to buy certain brands of products. Due to a large non-English speaking population in India, we observe a significant percentage of code-mix Hinglish search queries e.g., sasta atta. An intuitive approach to dealing with code-mix queries is to train an encoder-decoder model to translate the query to English to perform the search. However, the problem becomes non-trivial when the brand names themselves have Hinglish names and possibly have a literal English translation. In such queries, only the context (non-brand name) Hinglish words needs to be translated. In this paper, we propose a simple yet effective modification to the transformer training to preserve/correct Grocery brand names in the output while selectively translating the context words. To achieve this, we use an additional dataset of popular Grocery brand names. Brand names are added as tokens to the model vocabulary, and the token embeddings are randomly initialized. Further, we introduce a Brand loss in training the translation model. Brand loss is a cross entropy loss computed using a denoising auto-encoder objective with brand name data. We warm-start the training from a public pre-trained checkpoint (such as BART/T5) and further adapt it for query translation using the domain data. The proposed model is generic and can be used with English as well as code-mix Hinglish queries alleviating the need for language detection. To reduce the latency of the model for the production deployment, we use knowledge distillation and quantization. Experimental evaluation indicates that the proposed approach improves translation results by preserving/correcting English/Hinglish brand names. After positive results with A/B testing, the model is currently deployed in production.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="kulkarni-etal-2023-domain">
<titleInfo>
<title>Domain-specific transformer models for query translation</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mandar</namePart>
<namePart type="family">Kulkarni</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nikesh</namePart>
<namePart type="family">Garera</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anusua</namePart>
<namePart type="family">Trivedi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sunayana</namePart>
<namePart type="family">Sitaram</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Beata</namePart>
<namePart type="family">Beigman Klebanov</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jason</namePart>
<namePart type="given">D</namePart>
<namePart type="family">Williams</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Due to the democratization of e-commerce, many product companies are listing their goods for online shopping. For periodic buying within a domain such as Grocery, consumers are generally inclined to buy certain brands of products. Due to a large non-English speaking population in India, we observe a significant percentage of code-mix Hinglish search queries e.g., sasta atta. An intuitive approach to dealing with code-mix queries is to train an encoder-decoder model to translate the query to English to perform the search. However, the problem becomes non-trivial when the brand names themselves have Hinglish names and possibly have a literal English translation. In such queries, only the context (non-brand name) Hinglish words needs to be translated. In this paper, we propose a simple yet effective modification to the transformer training to preserve/correct Grocery brand names in the output while selectively translating the context words. To achieve this, we use an additional dataset of popular Grocery brand names. Brand names are added as tokens to the model vocabulary, and the token embeddings are randomly initialized. Further, we introduce a Brand loss in training the translation model. Brand loss is a cross entropy loss computed using a denoising auto-encoder objective with brand name data. We warm-start the training from a public pre-trained checkpoint (such as BART/T5) and further adapt it for query translation using the domain data. The proposed model is generic and can be used with English as well as code-mix Hinglish queries alleviating the need for language detection. To reduce the latency of the model for the production deployment, we use knowledge distillation and quantization. Experimental evaluation indicates that the proposed approach improves translation results by preserving/correcting English/Hinglish brand names. After positive results with A/B testing, the model is currently deployed in production.</abstract>
<identifier type="citekey">kulkarni-etal-2023-domain</identifier>
<identifier type="doi">10.18653/v1/2023.acl-industry.10</identifier>
<location>
<url>https://aclanthology.org/2023.acl-industry.10</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>89</start>
<end>95</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Domain-specific transformer models for query translation
%A Kulkarni, Mandar
%A Garera, Nikesh
%A Trivedi, Anusua
%Y Sitaram, Sunayana
%Y Beigman Klebanov, Beata
%Y Williams, Jason D.
%S Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track)
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F kulkarni-etal-2023-domain
%X Due to the democratization of e-commerce, many product companies are listing their goods for online shopping. For periodic buying within a domain such as Grocery, consumers are generally inclined to buy certain brands of products. Due to a large non-English speaking population in India, we observe a significant percentage of code-mix Hinglish search queries e.g., sasta atta. An intuitive approach to dealing with code-mix queries is to train an encoder-decoder model to translate the query to English to perform the search. However, the problem becomes non-trivial when the brand names themselves have Hinglish names and possibly have a literal English translation. In such queries, only the context (non-brand name) Hinglish words needs to be translated. In this paper, we propose a simple yet effective modification to the transformer training to preserve/correct Grocery brand names in the output while selectively translating the context words. To achieve this, we use an additional dataset of popular Grocery brand names. Brand names are added as tokens to the model vocabulary, and the token embeddings are randomly initialized. Further, we introduce a Brand loss in training the translation model. Brand loss is a cross entropy loss computed using a denoising auto-encoder objective with brand name data. We warm-start the training from a public pre-trained checkpoint (such as BART/T5) and further adapt it for query translation using the domain data. The proposed model is generic and can be used with English as well as code-mix Hinglish queries alleviating the need for language detection. To reduce the latency of the model for the production deployment, we use knowledge distillation and quantization. Experimental evaluation indicates that the proposed approach improves translation results by preserving/correcting English/Hinglish brand names. After positive results with A/B testing, the model is currently deployed in production.
%R 10.18653/v1/2023.acl-industry.10
%U https://aclanthology.org/2023.acl-industry.10
%U https://doi.org/10.18653/v1/2023.acl-industry.10
%P 89-95
Markdown (Informal)
[Domain-specific transformer models for query translation](https://aclanthology.org/2023.acl-industry.10) (Kulkarni et al., ACL 2023)
ACL
- Mandar Kulkarni, Nikesh Garera, and Anusua Trivedi. 2023. Domain-specific transformer models for query translation. In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 5: Industry Track), pages 89–95, Toronto, Canada. Association for Computational Linguistics.