Hierarchical corpus encoder: Fusing generative retrieval and dense indices
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916630762618880 |
|---|---|
| author | Chen, Tongfei Sharma, Ankita Pauls, Adam Van Durme, Benjamin |
| author_facet | Chen, Tongfei Sharma, Ankita Pauls, Adam Van Durme, Benjamin |
| contents | Generative retrieval employs sequence models for conditional generation of document IDs based on a query (DSI (Tay et al., 2022); NCI (Wang et al., 2022); inter alia). While this has led to improved performance in zero-shot retrieval, it is a challenge to support documents not seen during training. We identify the performance of generative retrieval lies in contrastive training between sibling nodes in a document hierarchy. This motivates our proposal, the hierarchical corpus encoder (HCE), which can be supported by traditional dense encoders. Our experiments show that HCE achieves superior results than generative retrieval models under both unsupervised zero-shot and supervised settings, while also allowing the easy addition and removal of documents to the index. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_18877 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Hierarchical corpus encoder: Fusing generative retrieval and dense indices Chen, Tongfei Sharma, Ankita Pauls, Adam Van Durme, Benjamin Information Retrieval Generative retrieval employs sequence models for conditional generation of document IDs based on a query (DSI (Tay et al., 2022); NCI (Wang et al., 2022); inter alia). While this has led to improved performance in zero-shot retrieval, it is a challenge to support documents not seen during training. We identify the performance of generative retrieval lies in contrastive training between sibling nodes in a document hierarchy. This motivates our proposal, the hierarchical corpus encoder (HCE), which can be supported by traditional dense encoders. Our experiments show that HCE achieves superior results than generative retrieval models under both unsupervised zero-shot and supervised settings, while also allowing the easy addition and removal of documents to the index. |
| title | Hierarchical corpus encoder: Fusing generative retrieval and dense indices |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2502.18877 |