Scaling Multilingual Semantic Search in Uber Eats Delivery
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912959896223744 |
|---|---|
| author | Ling, Bo Liu, Zheng Chen, Haoyang Nagar, Divya Yang, Luting Parsana, Mehul |
| author_facet | Ling, Bo Liu, Zheng Chen, Haoyang Nagar, Divya Yang, Luting Parsana, Mehul |
| contents | We present a production-oriented semantic retrieval system for Uber Eats that unifies retrieval across stores, dishes, and grocery/retail items. Our approach fine-tunes a Qwen2 two-tower base model using hundreds of millions of query-document interactions that were aggregated and anonymized pretraining. We train the model with a combination of InfoNCE on in-batch negatives and triplet-NCE loss on hard negatives, and we leverage Matryoshka Representation Learning (MRL) to serve multiple embedding sizes from a single model. Our system achieves substantial recall gains over a strong baseline across six markets and three verticals. This paper presents the end to end work including data curation, model architecture, large-scale training, and evaluation. We also share key insights and practical lessons for building a unified, multilingual, and multi-vertical retrieval system for consumer search. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_06586 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Scaling Multilingual Semantic Search in Uber Eats Delivery Ling, Bo Liu, Zheng Chen, Haoyang Nagar, Divya Yang, Luting Parsana, Mehul Information Retrieval H.3.3; I.2.7 We present a production-oriented semantic retrieval system for Uber Eats that unifies retrieval across stores, dishes, and grocery/retail items. Our approach fine-tunes a Qwen2 two-tower base model using hundreds of millions of query-document interactions that were aggregated and anonymized pretraining. We train the model with a combination of InfoNCE on in-batch negatives and triplet-NCE loss on hard negatives, and we leverage Matryoshka Representation Learning (MRL) to serve multiple embedding sizes from a single model. Our system achieves substantial recall gains over a strong baseline across six markets and three verticals. This paper presents the end to end work including data curation, model architecture, large-scale training, and evaluation. We also share key insights and practical lessons for building a unified, multilingual, and multi-vertical retrieval system for consumer search. |
| title | Scaling Multilingual Semantic Search in Uber Eats Delivery |
| topic | Information Retrieval H.3.3; I.2.7 |
| url | https://arxiv.org/abs/2603.06586 |