Reranking Passages with Coarse-to-Fine Neural Retriever Enhanced by List-Context Information

• URL: http://arxiv.org/abs/2308.12022v2
• Date: Thu, 21 Mar 2024 09:11:22 GMT
• Title: Reranking Passages with Coarse-to-Fine Neural Retriever Enhanced by List-Context Information
• Authors: Hongyin Zhu,
• Abstract summary: This paper presents a list-context attention mechanism to augment the passage representation by incorporating the list-context information from other candidates. The proposed coarse-to-fine (C2F) neural retriever addresses the out-of-memory limitation of the passage attention mechanism. It integrates the coarse and fine rankers into the joint optimization process, allowing for feedback between the two layers to update the model simultaneously.
• Score: 0.9463895540925061
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Passage reranking is a critical task in various applications, particularly when dealing with large volumes of documents. Existing neural architectures have limitations in retrieving the most relevant passage for a given question because the semantics of the segmented passages are often incomplete, and they typically match the question to each passage individually, rarely considering contextual information from other passages that could provide comparative and reference information. This paper presents a list-context attention mechanism to augment the passage representation by incorporating the list-context information from other candidates. The proposed coarse-to-fine (C2F) neural retriever addresses the out-of-memory limitation of the passage attention mechanism by dividing the list-context modeling process into two sub-processes with a cache policy learning algorithm, enabling the efficient encoding of context information from a large number of candidate answers. This method can be generally used to encode context information from any number of candidate answers in one pass. Different from most multi-stage information retrieval architectures, this model integrates the coarse and fine rankers into the joint optimization process, allowing for feedback between the two layers to update the model simultaneously. Experiments demonstrate the effectiveness of the proposed approach.

Related papers

• Leveraging Inter-Chunk Interactions for Enhanced Retrieval in Large Language Model-Based Question Answering [12.60063463163226]
  IIER captures the internal connections between document chunks by considering three types of interactions: structural, keyword, and semantic. It identifies multiple seed nodes based on the target question and iteratively searches for relevant chunks to gather supporting evidence. It refines the context and reasoning chain, aiding the large language model in reasoning and answer generation.
  arXiv  Detail & Related papers  (2024-08-06T02:39:55Z)
• Passage-specific Prompt Tuning for Passage Reranking in Question Answering with Large Language Models [11.716595438057997]
  We propose passage-specific prompt tuning for reranking in open-domain question answering (PSPT) PSPT is a parameter-efficient method that fine-tunes learnable passage-specific soft prompts. We conducted extensive experiments utilizing the Llama-2-chat-7B model across three publicly available open-domain question answering datasets.
  arXiv  Detail & Related papers  (2024-05-31T07:43:42Z)
• Multi-Granularity Information Interaction Framework for Incomplete Utterance Rewriting [32.05944198256814]
  Recent approaches in Incomplete Utterance Rewriting (IUR) fail to capture the source of important words. We propose a novel and effective multi-task information interaction framework including context selection, edit matrix construction, and relevance merging.
  arXiv  Detail & Related papers  (2023-12-19T08:43:02Z)
• Incorporating Relevance Feedback for Information-Seeking Retrieval using Few-Shot Document Re-Ranking [56.80065604034095]
  We introduce a kNN approach that re-ranks documents based on their similarity with the query and the documents the user considers relevant. To evaluate our different integration strategies, we transform four existing information retrieval datasets into the relevance feedback scenario.
  arXiv  Detail & Related papers  (2022-10-19T16:19:37Z)
• Contextual Fine-to-Coarse Distillation for Coarse-grained Response Selection in Open-Domain Conversations [48.046725390986595]
  We propose a Contextual Fine-to-Coarse (CFC) distilled model for coarse-grained response selection in open-domain conversations. To evaluate the performance of our proposed model, we construct two new datasets based on the Reddit comments dump and Twitter corpus.
  arXiv  Detail & Related papers  (2021-09-24T08:22:35Z)
• Tradeoffs in Sentence Selection Techniques for Open-Domain Question Answering [54.541952928070344]
  We describe two groups of models for sentence selection: QA-based approaches, which run a full-fledged QA system to identify answer candidates, and retrieval-based models, which find parts of each passage specifically related to each question. We show that very lightweight QA models can do well at this task, but retrieval-based models are faster still.
  arXiv  Detail & Related papers  (2020-09-18T23:39:15Z)
• ClarQ: A large-scale and diverse dataset for Clarification Question Generation [67.1162903046619]
  We devise a novel bootstrapping framework that assists in the creation of a diverse, large-scale dataset of clarification questions based on postcomments extracted from stackexchange. We quantitatively demonstrate the utility of the newly created dataset by applying it to the downstream task of question-answering. We release this dataset in order to foster research into the field of clarification question generation with the larger goal of enhancing dialog and question answering systems.
  arXiv  Detail & Related papers  (2020-06-10T17:56:50Z)
• Query Resolution for Conversational Search with Limited Supervision [63.131221660019776]
  We propose QuReTeC (Query Resolution by Term Classification), a neural query resolution model based on bidirectional transformers. We show that QuReTeC outperforms state-of-the-art models, and furthermore, that our distant supervision method can be used to substantially reduce the amount of human-curated data required to train QuReTeC.
  arXiv  Detail & Related papers  (2020-05-24T11:37:22Z)
• Multi-Stage Conversational Passage Retrieval: An Approach to Fusing Term Importance Estimation and Neural Query Rewriting [56.268862325167575]
  We tackle conversational passage retrieval (ConvPR) with query reformulation integrated into a multi-stage ad-hoc IR system. We propose two conversational query reformulation (CQR) methods: (1) term importance estimation and (2) neural query rewriting. For the former, we expand conversational queries using important terms extracted from the conversational context with frequency-based signals. For the latter, we reformulate conversational queries into natural, standalone, human-understandable queries with a pretrained sequence-tosequence model.
  arXiv  Detail & Related papers  (2020-05-05T14:30:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.