Neural Retrieval for Question Answering with Cross-Attention Supervised Data Augmentation

Yinfei Yang, Ning Jin, Kuo Lin, Mandy Guo, Daniel Cer


Abstract
Early fusion models with cross-attention have shown better-than-human performance on some question answer benchmarks, while it is a poor fit for retrieval since it prevents pre-computation of the answer representations. We present a supervised data mining method using an accurate early fusion model to improve the training of an efficient late fusion retrieval model. We first train an accurate classification model with cross-attention between questions and answers. The cross-attention model is then used to annotate additional passages in order to generate weighted training examples for a neural retrieval model. The resulting retrieval model with additional data significantly outperforms retrieval models directly trained with gold annotations on Precision at N (P@N) and Mean Reciprocal Rank (MRR).
Anthology ID:
2021.acl-short.35
Volume:
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)
Month:
August
Year:
2021
Address:
Online
Venues:
ACL | IJCNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
263–268
Language:
URL:
https://aclanthology.org/2021.acl-short.35
DOI:
10.18653/v1/2021.acl-short.35
Bibkey:
Cite (ACL):
Yinfei Yang, Ning Jin, Kuo Lin, Mandy Guo, and Daniel Cer. 2021. Neural Retrieval for Question Answering with Cross-Attention Supervised Data Augmentation. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 263–268, Online. Association for Computational Linguistics.
Cite (Informal):
Neural Retrieval for Question Answering with Cross-Attention Supervised Data Augmentation (Yang et al., ACL 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.acl-short.35.pdf
Data
ReQASQuAD