Computer Science > Artificial Intelligence
[Submitted on 6 Apr 2023 (v1), revised 11 Apr 2023 (this version, v3), latest version 1 Feb 2024 (v6)]
Title:Revisiting Dense Retrieval with Unanswerable Counterfactuals
View PDFAbstract:The retriever-reader framework is popular for open-domain question answering (ODQA), where a retriever samples for the reader a set of relevant candidate passages from a large corpus. A key assumption behind this method is that high relevance scores from the retriever likely indicate high answerability from the reader, which implies a high probability that the retrieved passages contain answers to a given question. In this work, we empirically dispel this belief and observe that recent dense retrieval models based on DPR often rank unanswerable counterfactual passages higher than their answerable original passages. To address such answer-unawareness in dense retrievers, we seek to use counterfactual samples as additional training resources to better synchronize the relevance measurement of DPR with the answerability of question-passage pairs. Specifically, we present counterfactually-Pivoting Contrastive Learning (PiCL), a novel representation learning approach for passage retrieval that leverages counterfactual samples as pivots between positive and negative samples in their learned embedding space. We incorporate PiCL into the retriever training to show the effectiveness of PiCL on ODQA benchmarks and the robustness of the learned models.
Submission history
From: Yongho Song [view email][v1] Thu, 6 Apr 2023 12:42:37 UTC (6,583 KB)
[v2] Mon, 10 Apr 2023 05:20:12 UTC (6,583 KB)
[v3] Tue, 11 Apr 2023 08:15:15 UTC (6,583 KB)
[v4] Wed, 12 Apr 2023 11:09:59 UTC (6,730 KB)
[v5] Thu, 4 May 2023 06:40:20 UTC (6,736 KB)
[v6] Thu, 1 Feb 2024 17:36:39 UTC (9,046 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.