Improving Conversational Passage Re-ranking with View Ensemble

Abstract

This paper presents ConvRerank, a conversational passage re-ranker that employs a newly developed pseudo-labeling approach. Our proposed view-ensemble method enhances the quality of pseudo-labeled data, thus improving the fine-tuning of ConvRerank. Our experimental evaluation on benchmark datasets shows that combining ConvRerank with a conversational dense retriever in a cascaded manner achieves a good balance between effectiveness and efficiency. Compared to baseline methods, our cascaded pipeline demonstrates lower latency and higher top-ranking effectiveness. Furthermore, the in-depth analysis confirms the potential of our approach to improving the effectiveness of conversational search.

SIGIR 2023

References in corpus (6)

• Multi-Stage Document Ranking with BERT
• Query Resolution for Conversational Search with Limited Supervision
• Open-Retrieval Conversational Question Answering
• Few-Shot Conversational Dense Retrieval
• Conversational Question Reformulation via Sequence-to-Sequence Architectures and Pretrained Language Models
• Zero-shot Query Contextualization for Conversational Search