Contextualized query expansion via unsupervised chunk selection for text retrieval

作者：

Highlights：

• A BERT-based query expansion (QE) model to identify relevant information from text.

• Novel QE components to better trade-off efficiency against effectiveness.

• Evaluation on two standard TREC test collections demonstrates superior performance.

• Analysis provides insights on how to fine-tune BERT ranker for long documents.

摘要

•A BERT-based query expansion (QE) model to identify relevant information from text.•Novel QE components to better trade-off efficiency against effectiveness.•Evaluation on two standard TREC test collections demonstrates superior performance.•Analysis provides insights on how to fine-tune BERT ranker for long documents.

论文关键词：00-01,99-00,Contextualized query expansion,Document re-ranking,Pre-trained text representations

论文评审过程：Received 29 March 2021, Revised 21 June 2021, Accepted 25 June 2021, Available online 9 July 2021, Version of Record 9 July 2021.

论文官网地址：https://doi.org/10.1016/j.ipm.2021.102672