HOTPOTQA

Predicting Difficulty and Discrimination of Natural Language Questions (2022.acl-short)

Challenge:	a persistent gap remains between Recall@10 and Recall @50 across datasets .
Approach:	They evaluate embedding model comparison, Reciprocal Rank Fusion and embedded concatenation techniques to improve retrieval quality.
Outcome:	The proposed methods outperform traditional cross-encoders in identifying high-relevance passages.

Copied to clipboard

Matthew Byrd, Shashank Srivastava

Challenge:	Item Response Theory (IRT) has been used to numerically characterize question difficulty and discrimination for human subjects in domains including cognitive psychology and education.
Approach:	They explore the relationship between difficulty and discrimination in question-answering contexts by using IRT to characterize item difficulty and item discrimination.
Outcome:	The proposed models can predict difficulty and discrimination parameters for new questions and explain them with features of questions, answers, and associated contexts.

Question Decomposition for Retrieval-Augmented Generation (2025.acl-srw)

Copied to clipboard

Paul J. L. Ammann, Jonas Golde, Alan Akbik

Test-Time Strategies for More Efficient and Accurate Agentic RAG (2026.acl-srw)

Challenge:	Retrieval-augmented generation (RAG) is effective for question answering tasks . multi-hop questions, such as "Which company among NVIDIA, Apple, and Google made the biggest profit in 2023?" challenge RAG because relevant facts are often distributed across multiple documents .
Approach:	They propose a pipeline that incorporates question decomposition to ground large language models in verifiable external sources.
Outcome:	The proposed approach improves retrieval and answer accuracy over standard RAG . multi-hop questions often require multiple documents to support the model .

Copied to clipboard

Abhinav Sharma, Brian Zhang, Deepti Guntur, Zhiyang Zuo, Shreyas Chaudhari, Wenlong Zhao, Franck Dernoncourt, Puneet Mathur, Ryan A. Rossi, Nedim Lipka

How Well Do Multi-hop Reading Comprehension Models Understand Date Information? (2022.aacl-short)

Challenge:	Retrieval-Augmented Generation (RAG) systems face challenges with complex, multi-hop questions.
Approach:	They propose to integrate contextualization module and de-duplication module to improve the accuracy of retrieved documents and to reduce the number of turns by 10.5%.
Outcome:	The proposed approach achieves a 5.6% increase in EM score and reduces the average number of turns by 10.5% compared to the baseline.

Copied to clipboard

Xanh Ho, Saku Sugawara, Akiko Aizawa

Detecting Overflow in Compressed Token Representations for Retrieval-Augmented Generation (2026.eacl-srw)

Challenge:	Existing multi-hop reading comprehension datasets have reasoning shortcuts that can be used to answer comparison questions without performing multi- hop reasoning.
Approach:	They propose a dataset with three probing tasks in addition to the main question . they then evaluate the model's ability to understand date information .
Outcome:	The proposed model performs well in date comparison and number subtraction tasks.

Copied to clipboard

Julia Belikova, Danila Rozhevskii, Dennis Svirin, Konstantin Polev, Alexander Panchenko

Representation Learning

Challenge:	Efficient long-context processing remains a challenge for large language models (LLMs) however, the limits of compressibility remain underexplored.
Approach:	They propose a method to characterize and detect token overflow in xRAG soft-compression by mapping long contexts into dense vectors that can be directly consumed by the model.
Outcome:	The proposed method identifies token overflow with query-agnostic saturation statistics but lacks the capability to detect it.

HiGraAgent: Dual-Agent Adaptive Reasoning over Hierarchical Knowledge Graph for Open Domain Multi-hop Question Answering (2026.findings-eacl)

Copied to clipboard

Hung Luu, Long S. T. Nguyen, Trung Pham, Hieu Pham, Tho Quan

Structured Data in NLP

Challenge:	Existing approaches to multi-hop question answering lack a robust and flexible approach to QA . prior work showed compositionality gap persists even for Large Language Models .
Approach:	They propose a framework that unifies graph-based retrieval with adaptive reasoning . HiGraAgent uses a hierarchical knowledge Graph with entity alignment .
Outcome:	The proposed framework outperforms the strongest graph-based method on hotpotQA, 2WikiMultihopQA, and MuSiQue.

Modeling Multi-hop Question Answering as Single Sequence Prediction (2022.acl-long)

Copied to clipboard

Semih Yavuz, Kazuma Hashimoto, Yingbo Zhou, Nitish Shirish Keskar, Caiming Xiong

RealFormer: Transformer Likes Residual Attention (2021.findings-acl)

Challenge:	Existing generative question answering models that leverage passage retrieval with a pre-trained transformer are not effective for multihop QA.
Approach:	They propose a generative approach that explicitly models the reasoning process to resolve the answer for multi-hop questions by encoding cross-passage interactions.
Outcome:	The proposed model improves on two multi-hop QA datasets and is interpretable.

Copied to clipboard

Ruining He, Anirudh Ravula, Bhargav Kanagal, Joshua Ainslie

Utility-Focused LLM Annotation for Retrieval and Retrieval-Augmented Generation (2025.emnlp-main)

Challenge:	Existing techniques to create Residual Attention Layer Transformer networks outperform the canonical Transformer on a wide spectrum of tasks.
Approach:	They propose a technique to create Residual Attention Layer Transformer networks that outperform the canonical Transformer on a wide spectrum of tasks.
Outcome:	The proposed technique outperforms the canonical Transformer on a wide spectrum of tasks including Masked Language Modeling, GLUE, SQUAD, Neural Machine Translation, WikiHop, HotpotQA, Natural Questions, and OpenKP.

Copied to clipboard

Hengran Zhang, Minghao Tang, Keping Bi, Jiafeng Guo, Shihao Liu, Daiting Shi, Dawei Yin, Xueqi Cheng

Chain-of-Skills: A Configurable Model for Open-Domain Question Answering (2023.acl-long)

Challenge:	Existing studies on large language models for document utility annotations have shown that they improve retrieval performance and RAG outcomes compared to models trained on human annotations.
Approach:	They propose a model that maximizes their summed marginal likelihood to annotate document utility on multiple positive samples per query.
Outcome:	The proposed model maximizes the marginal likelihood of multiple positive samples per query.

Copied to clipboard

Kaixin Ma, Hao Cheng, Yu Zhang, Xiaodong Liu, Eric Nyberg, Jianfeng Gao

End-to-End Beam Retrieval for Multi-Hop Question Answering (2024.naacl-long)

Challenge:	Using customized retrieval models, model transferability and scalability are limited.
Approach:	They propose a modular retrieval model where individual modules correspond to key skills that can be reused across datasets.
Outcome:	The proposed model outperforms self-supervised retrievers in zero-shot evaluations and achieves state-of-the-art fine-tuned retrieval performance on NQ, HotpotQA and OTT-QA.

Copied to clipboard

Jiahao Zhang, Haiyang Zhang, Dongmei Zhang, Liu Yong, Shen Huang

EXPLAIN: Enhancing Retrieval-Augmented Generation with Entity Summary (2025.acl-industry)

Challenge:	Existing beam retrieval frameworks for multi-hop question answering were customized for two-hop questions and were poorly supervised.
Approach:	They propose an end-to-end beam retrieval framework for multi-hop question answering . they combine an encoder and two classification heads to optimize the retrieval process .
Outcome:	The proposed framework improves on MuSiQue-Ans and surpasses all previous retrievers on HotpotQA and achieves 99.9% precision on 2WikiMultiHopQA.

Copied to clipboard

Yaozhen Liang, Xiao Liu, Jiajun Yu, Zhouhua Fang, Qunsheng Zou, Linghan Zheng, Yong Li, Zhiwei Liu, Haishuai Wang

Explainability & Interpretability in NLP

Challenge:	Existing document question answering methods reduce inference costs and input tokens.
Approach:	They propose a retrieval-augmented generation method that automatically extracts useful entities and generates summaries from documents.
Outcome:	The proposed method surpasses baseline retrieval-augmented generation (RAG) and long-context question answering (LC) methods achieve higher accuracy by processing entire documents, but at the cost of increased computational Corresponding authors.

Learning to Decompose: Hypothetical Question Decomposition Based on Comparable Texts (2022.emnlp-main)

Copied to clipboard

Ben Zhou, Kyle Richardson, Xiaodong Yu, Dan Roth

ASRank: Zero-Shot Re-Ranking with Answer Scent for Document Retrieval (2025.findings-naacl)

Challenge:	Existing approaches to end-to-end questionanswering assume that pre-trained language can decompose complex tasks into more straightforward sub-tasks.
Approach:	They propose to use distant supervision to train decomposition-based transformers for large-scale parallel news.
Outcome:	The proposed model improves on semantic parsing and on hotpotQA and strategyQA datasets by 20% to 30%.

Copied to clipboard

Abdelrahman Abdallah, Jamshid Mozafari, Bhawna Piryani, Adam Jatowt

Momoka-RAG: MCTS-Organized Mapping of Knowledge Associations for Long-Document Retrieval Augmented Generation (2026.findings-acl)

Challenge:	Retrieval-Augmented Generation models fail to rank the most relevant documents at the top . conventional retrieval methods fail to find the most important documents .
Approach:	They propose a new method for scoring retrieved documents using zero-shot answer scent based on a pre-trained large language model to compute the likelihood of document-derived answers aligning with the answer scent.
Outcome:	The proposed method improves top-1 retrieval accuracy on NQ, TriviaQA, WebQA, ArchivalQA, HotpotQA, and Entity Questions.

Copied to clipboard

Wenyu Tao, Xiaofen Xing, Zeliang Li, Xiangmin Xu

Context-Aware Answer Extraction in Question Answering (2020.emnlp-main)

Challenge:	Existing frameworks that rely on fixed-length chunking are unsuitable for long-document tasks due to their passive and mechanical approach to knowledge structure.
Approach:	They propose a framework that utilizes Monte Carlo Tree Search to proactively uncover connections among chunks and construct optimal semantic information paths with the objective of completing semantic relationships.
Outcome:	The proposed framework achieves higher precision while maintaining competitive recall compared to other RAG frameworks.

Copied to clipboard

Yeon Seonwoo, Ji-Hoon Kim, Jung-Woo Ha, Alice Oh

When and How to Augment Your Input: Question Routing Helps Balance the Accuracy and Efficiency of Large Language Models (2025.findings-naacl)

Challenge:	Extractive QA models have shown promising performance in predicting the correct answer to a given question.
Approach:	They propose a BLANC-based context prediction task that learns the context prediction tasks.
Outcome:	The proposed model outperforms the state-of-the-art models on reading comprehension and hotpotQA.

Copied to clipboard

Shufan Chen, He Zheng, Lei Cui

AutoPlan: Automatic Planning of Interactive Decision-Making Tasks With Large Language Models (2023.findings-emnlp)

Challenge:	augmented generation of knowledge-based long-tail questions can be useful for large language models, but can cause significant latency.
Approach:	They propose an adaptive question routing framework that uses a query router to augment input to the right time.
Outcome:	The proposed framework surpasses existing approaches in accuracy and efficiency on benchmarks such as AmbigNQ, HotpotQA, MMLU-STEM, and PopQA.

Copied to clipboard

Siqi Ouyang, Lei Li

KnowAgent: Knowledge-Augmented Planning for LLM-Based Agents (2025.findings-naacl)

Challenge:	Existing methods for making decisions in grounded environments require costly gradient computation or lengthy in-context demonstrations.
Approach:	They propose an approach to guide LLM-based agents to accomplish interactive decision-making tasks by using an LLM prompt and a task-solving plan.
Outcome:	The proposed approach outperforms human-written demonstrations on ALFWorld and HotpotQA by 8%.

Copied to clipboard

Yuqi Zhu, Shuofei Qiao, Yixin Ou, Shumin Deng, Shiwei Lyu, Yue Shen, Lei Liang, Jinjie Gu, Huajun Chen, Ningyu Zhang

Long Context Question Answering via Supervised Contrastive Learning (2022.naacl-main)

Challenge:	Large Language Models (LLMs) fail to effectively guide the planning trajectories during task solving and result in planning hallucinations.
Approach:	They propose a novel approach to enhance the planning capabilities of large language models by incorporating explicit action knowledge.
Outcome:	The proposed approach can achieve comparable or superior performance to existing baselines on HotpotQA and ALFWorld.

Copied to clipboard

Avi Caciularu, Ido Dagan, Jacob Goldberger, Arman Cohan

Multi-Hop Paragraph Retrieval for Open-Domain Question Answering (P19-1)

Challenge:	Long-context question answering tasks often require identifying evidence spans (e.g., sentences) prior work showed that jointly training models to perform evidence extraction and question answering is important for achieving high performance.
Approach:	They propose a method for equipping long-context QA models with an additional sequence-level objective for better identification of the supporting evidence.
Outcome:	The proposed method exhibits consistent improvements on three different strong long-context transformer models, across two challenging question answering benchmarks – HotpotQA and QAsper.

Copied to clipboard

Yair Feldman, Ran El-Yaniv

Answering while Summarizing: Multi-task Learning for Multi-hop QA with Evidence Extraction (P19-1)

Challenge:	Existing methods for textual question answering are capable of outperforming humans on certain tasks.
Approach:	They propose a method for retrieving multiple supporting paragraphs from a large knowledge base.
Outcome:	The proposed method achieves state-of-the-art over two well-known datasets, SQuAD-Open and HotpotQA, which serve as benchmarks for the proposed method.

Copied to clipboard

Kosuke Nishida, Kyosuke Nishida, Masaaki Nagata, Atsushi Otsuka, Itsumi Saito, Hisako Asano, Junji Tomita

Information Extraction & Text Mining

Challenge:	Question answering (QA) using textual sources for purposes such as reading comprehension has attracted much attention.
Approach:	They propose a Query Focused Extractor model for evidence extraction and multi-task learning with the QA model.
Outcome:	The proposed model achieves state-of-the-art evidence extraction score on hotpotQA and FEVER, which is a recognizing textual entailment task on a large textual database.

Span Selection Pre-training for Question Answering (2020.acl-main)

Copied to clipboard

Michael Glass, Alfio Gliozzo, Rishav Chakravarti, Anthony Ferritto, Lin Pan, G P Shrivatsa Bhargav, Dinesh Garg, Avi Sil

Can Small Language Models Help Large Language Models Reason Better?: LM-Guided Chain-of-Thought (2024.lrec-main)

Challenge:	Pre-trained BERTs provide large gains across many language understanding tasks, achieving a new state-of-the-art (SOTA).
Approach:	They propose a new pre-training task inspired by reading comprehension to better align the pre- training from memorization to understanding.
Outcome:	The proposed model outperforms BERT-BASE and BERT LARGE on a new dataset and improves answer prediction F1 by 4 points and supporting fact prediction F1.

Copied to clipboard

Jooyoung Lee, Fan Yang, Thanh Tran, Qian Hu, Emre Barut, Kai-Wei Chang

HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (D18-1)

Challenge:	Existing frameworks for guiding a language model in reasoning tasks are limited by their tendency to generate low-quality rationales that are repetitive and vacuous.
Approach:	They propose a framework that leverages a lightweight language model for guiding a black-box large LM in reasoning tasks.
Outcome:	The proposed framework outperforms baselines in answer prediction accuracy.

Copied to clipboard

Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, Christopher D. Manning

Answering Complex Open-domain Questions Through Iterative Query Generation (D19-1)

Challenge:	Existing question answering (QA) datasets fail to train QA systems to perform complex reasoning and provide explanations for answers.
Approach:	They propose a new dataset with 113k Wikipedia-based question-answer pairs with four key features: (1) the questions require finding and reasoning over multiple supporting documents to answer; (2) the questions are diverse and not constrained to any pre-existing knowledge bases or knowledge schemas; (3) the questions provide sentence-level supporting facts required for reasoning; and (4) a type of factoid comparison questions to test QA systems’ ability to extract relevant facts and perform necessary comparison.
Outcome:	The proposed dataset has 113k Wikipedia-based question-answer pairs and four key features that make it challenging for the latest QA systems.

Copied to clipboard

Peng Qi, Xiaowen Lin, Leo Mehr, Zijian Wang, Christopher D. Manning

On-Policy Self-Alignment with Fine-grained Knowledge Feedback for Hallucination Mitigation (2025.findings-acl)

Challenge:	Currently, one-step retrieve-and-read question answering systems cannot answer such questions because they rarely contain retrievable clues about the missing entity.
Approach:	They propose a multi-step approach to retrieve relevant content with the question, then reading the paragraphs returned by the information retrieval component to arrive at the final answer.
Outcome:	The proposed model outperforms the best previously published model despite not using pretrained language models such as BERT.

Copied to clipboard

Xueru Wen, Jie Lou, Xinyu Lu, Yuqiu Ji, Xinyan Guan, Yaojie Lu, Hongyu Lin, Ben He, Xianpei Han, Debing Zhang, Le Sun

HIRAG: Hierarchical-Thought Instruction-Tuning Retrieval-Augmented Generation (2025.findings-emnlp)

Challenge:	Large language models exhibit behavior that deviates from the boundaries of their knowledge during response generation.
Approach:	They propose a framework that allows large language models to explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals.
Outcome:	The proposed framework enables LLMs to explore their knowledge boundaries and self-correct generation behavior through fine-grained feedback signals.

Copied to clipboard

Yihan Jiao, Zhehao Tan, Dan Yang, Duolin Sun, Jie Feng, Yue Shen, Jian Wang, Peng Wei

HPE: Answering Complex Questions over Text by Hybrid Question Parsing and Execution (2023.findings-emnlp)

Challenge:	In-depth research on the specific capabilities needed by the RAG generation model is lacking, leading to inconsistent document quality and retrieval system imperfections.
Approach:	They propose that RAG models should possess three progressively hierarchical abilities: (1) Filtering: the ability to select relevant information; (2) Combination: the capability to combine semantic information across paragraphs; (3) RAG-specific reasoning: the capacity to further process external knowledge using internal knowledge.
Outcome:	Experiments show that the proposed method significantly improves the model’s open-book examination capability on datasets such as RGB, PopQA, MuSiQue, HotpotQA, and PubmedQA.

Copied to clipboard

Ye Liu, Semih Yavuz, Rui Meng, Dragomir Radev, Caiming Xiong, Shafiq Joty, Yingbo Zhou

BRIEF: Bridging Retrieval and Inference for Multi-hop Reasoning via Compression (2025.findings-naacl)

Challenge:	End-to-end neural networks excel at answering natural language questions but fail on complex ones . a proposed framework for question parsing and execution on textual QA is designed to combine the strengths of neural and symbolic methods.
Approach:	They propose a framework for question parsing and execution on textual QA . they parse questions into an intermediate representation and use deterministic rules to translate them .
Outcome:	The proposed framework outperforms existing methods in supervised, few-shot, and zero-shot settings while preserving its underlying reasoning process.

Copied to clipboard

Yuankai Li, Jia-Chen Gu, Di Wu, Kai-Wei Chang, Nanyun Peng

Adapting LLM Agents with Universal Communication Feedback (2025.findings-naacl)

Challenge:	Existing approaches to augment language models with external knowledge but they are limited by static nature of pre-training data.
Approach:	They propose a lightweight approach that compresses retrieved documents into highly dense textual summaries to integrate into in-context RAG.
Outcome:	The proposed approach reduces latency and costs while achieving high performance in open-domain questions.

Copied to clipboard

Kuan Wang, Yadong Lu, Michael Santacroce, Yeyun Gong, Chao Zhang, Yelong Shen

Challenge:	Recent advances in large language models (LLMs) have demonstrated potential for LLM agents.
Approach:	They propose a universal buffer and iterative pipeline to store feedback and itersative pipelines to enable LLM agents to explore and update their policy in an environment.
Outcome:	The proposed approach outperforms supervised instruction fine-tuning baselines on four datasets.

Gumbel Reranking: Differentiable End-to-End Reranker Optimization (2025.acl-long)

Copied to clipboard

Siyuan Huang, Zhiyuan Ma, Jintao Du, Changhua Meng, Weiqiang Wang, Jingwen Leng, Minyi Guo, Zhouhan Lin

Wikontic: Constructing Wikidata-Aligned, Ontology-Aware Knowledge Graphs with Large Language Models (2026.eacl-long)

Challenge:	Existing distillation-based approaches suffer from training-inference misalignment and fail to capture interdependencies among candidate documents.
Approach:	They propose a method to optimize rerankers by learning a stochastic, document-wise Top-k attention mask using the Gumbel Trick and Relaxed Top-K Sampling.
Outcome:	The proposed framework minimizes the overall language loss and improves recall on hotpotQA.

Copied to clipboard

Alla Chepurova, Aydar Bulatov, Mikhail Burtsev, Yuri Kuratov

Towards Uncertainty-Aware Language Agent (2024.findings-acl)

Challenge:	Knowledge graphs provide structured, verifiable grounding for large language models . current LLMs use KGs as auxiliary structures for text retrieval .
Approach:	They propose a pipeline that constructs KGs from open-domain texts using triplets and qualifiers.
Outcome:	The proposed pipeline outperforms existing methods in retrieval-augmented generation.

Copied to clipboard

Jiuzhou Han, Wray Buntine, Ehsan Shareghi

Challenge:	Existing Language Agents neglect the notion of uncertainty during interactions with external worlds.
Approach:	They propose a framework that orchestrates the interaction between the agent and the external world using uncertainty quantification.
Outcome:	The proposed framework improves performance on 3 representative tasks and lowers reliance on external world.

Understanding Dataset Design Choices for Multi-hop Reasoning (N19-1)

Copied to clipboard

Jifan Chen, Greg Durrett

Compositional Questions Do Not Necessitate Multi-hop Reasoning (P19-1)

Challenge:	Existing datasets that explicitly focus on multi-hop reasoning are lacking in learning multi-tasking.
Approach:	They propose to use sentence-factored models to solve multi-hop question answering tasks . they find spurious correlations in unmasked versions of WikiHop and HotpotQA .
Outcome:	The proposed datasets are used to test models on multi-hop question answering tasks.

Copied to clipboard

Sewon Min, Eric Wallace, Sameer Singh, Matt Gardner, Hannaneh Hajishirzi, Luke Zettlemoyer

Self-Assembling Modular Networks for Interpretable Multi-Hop Reasoning (D19-1)

Challenge:	a single-hop reasoning model can solve much more of the dataset than previously thought.
Approach:	They propose a single-hop BERT-based RC model that achieves 67 F1 . they propose an evaluation setting where humans are not shown all paragraphs .
Outcome:	The proposed model achieves 67 F1—comparable to state-of-the-art multi-hop models.

Copied to clipboard

Yichen Jiang, Mohit Bansal

From Confidence to Collapse in LLM Factual Robustness (2025.findings-emnlp)

Challenge:	Existing models for multi-hop question answering require multiple pieces of evidence scattered in a given context.
Approach:	They propose an interpretable, controller-based self-assembling Neural Modular Network for multi-hop reasoning . their model can softly decompose a multi-step question into multiple single-hop sub-questions .
Outcome:	The proposed model improves on the static, single-hop model on regular and adversarial evaluations.

Copied to clipboard

Alina Fastowski, Bardh Prenkaj, Gjergji Kasneci

Robustness in NLP

Challenge:	Existing evaluation methods focus on performance-based metrics, often investigating from the perspective of prompt perturbations, which captures only the externally triggered side of knowledge robustness.
Approach:	They propose a method to measure factual robustness from the perspective of the generation process by analyzing token distribution entropy and temperature scaling sensitivity.
Outcome:	The proposed method measures factual robustness from the perspective of the generation process and entropy and temperature scaling sensitivity.

CQG: A Simple and Effective Controlled Generation Framework for Multi-hop Question Generation (2022.acl-long)

Copied to clipboard

Zichu Fei, Qi Zhang, Tao Gui, Di Liang, Sirui Wang, Wei Wu, Xuanjing Huang

Text Generation

Challenge:	Current models can not ensure the complexity of generated questions, so they may generate shallow questions that can be answered without multi-hop reasoning.
Approach:	They propose a controlled framework to generate multi-hop questions that contain key entities in multi- hop reasoning chains and a novel Transformer-based decoder to guarantee that key entities appear in the questions.
Outcome:	The proposed model outperforms the state-of-the-art model 25% on HotpotQA.

Robustifying Multi-hop QA through Pseudo-Evidentiality Training (2021.acl-long)

Copied to clipboard

Kyungjae Lee, Seung-won Hwang, Sang-eun Han, Dohyeon Lee

Dr3: Ask Large Language Models Not to Give Off-Topic Answers in Open Domain Multi-Hop Question Answering (2024.lrec-main)

Challenge:	Existing approaches to robustify multi-hop question answering models require expensive annotations.
Approach:	They propose a method to supervise answers with right reasoning chains without annotations . they compare answers confidence with and without evidence sentences to generate "pseudo-evidentiality" annotations.
Outcome:	The proposed model is accurate and robust in multi-hop reasoning.

Copied to clipboard

Yuan Gao, Yiheng Zhu, Yuanbin Cao, Yinzhi Zhou, Zhen Wu, Yujie Chen, Shenglan Wu, Haoyuan Hu, Xinyu Dai

MASTER: A Multi-Agent System with LLM Specialized MCTS (2025.naacl-long)

Challenge:	Open Domain Multi-Hop Question Answering (ODMHQA) is one of the most challenging tasks in Natural Language Processing (NLP)
Approach:	They propose a mechanism that leverages the intrinsic capabilities of Large Language Models to judge whether the generated answers are off-topic.
Outcome:	The proposed method reduces the occurrence of off-topic answers by nearly 13%, improving the performance in Exact Match (EM) by nearly 3% compared to the baseline method without the Dr3 mechanism.

Copied to clipboard

Bingzheng Gan, Yufan Zhao, Tianyi Zhang, Jing Huang, Li Yusu, Shu Xian Teo, Changwang Zhang, Wei Shi

Challenge:	Large Language Models (LLMs) are increasingly being explored for problem-solving tasks . their strategic planning capability is often viewed with skepticism due to their limited planning capabilities.
Approach:	They propose a framework that coordinates agent recruitment and communication through LLM specialized MCTS.
Outcome:	The proposed framework achieves 76% accuracy on HotpotQA and 80% on WebShop . it relies on extensive sampling simulations to approximate the true reward distribution .

A STEP towards Interpretable Multi-Hop Reasoning:Bridge Phrase Identification and Query Expansion (2022.lrec-1)

Copied to clipboard

Fan Luo, Mihai Surdeanu

Mitigating False-Negative Contexts in Multi-document Question Answering with Retrieval Marginalization (2021.emnlp-main)

Challenge:	Identifying bridge phrases remains one of the challenges for multi-hop question answering .
Approach:	They propose an unsupervised method for the identification of bridge phrases in multi-hop question answering . they construct a graph of noun phrases from the question and available context .
Outcome:	The proposed method improves all downstream components in a multi-hop QA system.

Copied to clipboard

Ansong Ni, Matt Gardner, Pradeep Dasigi

Logic-Guided Data Augmentation and Regularization for Consistent Question Answering (2020.acl-main)

Challenge:	Question Answering models typically use retrieval and reasoning components to identify relevant information for reasoning.
Approach:	They propose a retrieval parameterization method that marginalizes unanswerable queries . they show that marginalization allows a model to mitigate false negatives in annotations .
Outcome:	The proposed model improves on two multi-document question answering datasets and shows that marginalization improves performance.

Copied to clipboard

Akari Asai, Hannaneh Hajishirzi

NSF-CoT: Neuro-Symbolic Formal Verification of Chain-of-Thought Faithfulness in Contextual Question Answering (2026.findings-acl)

Challenge:	Recent studies have focused on improving the consistency of comparison questions . current models show inconsistent comparison predictions due to small datasets .
Approach:	They propose a method that integrates logic rules and neural models to improve the accuracy of comparison questions by enhancing labeled training data.
Outcome:	The proposed method improves state-of-the-art models on WIQA and QuaRel and reduces consistency violations by 58% on HotpotQA.

Copied to clipboard

Vishal Pramanik, Maisha Maliha, Nathaniel D. Bastian, Alvaro Velasquez, Susmit Jha, Sumit Kumar Jha

Multi-Hop Question Generation via Dual-Perspective Keyword Guidance (2025.findings-acl)

Challenge:	Existing faithfulness checks only test whether changing the written chain changes the answer, without verifying whether the steps are truly supported by the given evidence.
Approach:	They propose a method that checks CoT faithfulness step by step for contextual question answering.
Outcome:	The proposed method outperforms causal mediation, perturbation probes, and behavioral monitoring in OpenBookQA, QASC, and HotpotQA.

Copied to clipboard

Maodong Li, Longyin Zhang, Fang Kong

Uncertainty Quantification of Large Language Models through Multiple Uncertainty Sources (2026.findings-acl)

Challenge:	Existing work fails to fully utilize the guiding potential of keywords and neglect to differentiate the distinct roles of question-specific and document-specific keywords.
Approach:	They propose a dual-perspective keyword-guided framework that integrates question and document keywords into the multi-hop question generation process.
Outcome:	The proposed framework integrates question and document keywords into the multi-hop question generation process.

Copied to clipboard

Tiejin Chen, Xiaoou Liu, Longchao Da, Jia Chen, Evangelos E. Papalexakis, Hua Wei

Momentum Posterior Regularization for Multi-hop Dense Retrieval (2025.coling-main)

Challenge:	Existing methods for uncertainty quantification fail to capture multifaceted nature of natural language generation.
Approach:	They propose a multi-resource Uncertainty Quantification framework that integrates heterogeneous uncertainty signals into a unified measure.
Outcome:	The proposed framework outperforms existing methods on CoQA, NQ_Open, and HotpotQA.

Copied to clipboard

Zehua Xia, Yuyang Wu, Yiyun Xia, Cam Tu Nguyen

LinkBERT: Pretraining Language Models with Document Links (2022.acl-long)

Challenge:	Current methods for knowledge distillation in one-time retrieval are ineffective for multi-hop QA . posterior information is often defined as the response, which may not connect to the query without intermediate retrieval .
Approach:	They propose to distill knowledge from a posterior retrieval into a prior retrieval for multi-hop QA . they propose to use momentum moving average method to update posterior information along with prior retrievals .
Outcome:	Experiments on HotpotQA and StrategyQA show that MoPo outperforms baselines in retrieval and downstream QA tasks.

Copied to clipboard

Michihiro Yasunaga, Jure Leskovec, Percy Liang

Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions (2023.acl-long)

Challenge:	Existing language model pretraining methods do not capture dependencies or knowledge that span across documents.
Approach:	They propose a language model pretraining method that leverages links between documents . they use masked language modeling and document relation prediction to model LMs .
Outcome:	The proposed method outperforms existing methods on downstream tasks across two domains.

Copied to clipboard

Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, Ashish Sabharwal

Generative Context Pair Selection for Multi-hop Question Answering (2021.emnlp-main)

Challenge:	Large language models generate natural language reasoning steps or Chains-of-Thoughts when prompted appropriately.
Approach:	They propose a new approach that interleaves retrieval with steps (sentences) in a CoT and uses retrieved results to improve CoT.
Outcome:	The proposed approach improves retrieval and downstream QA significantly on four datasets.

Copied to clipboard

Dheeru Dua, Cicero Nogueira dos Santos, Patrick Ng, Ben Athiwaratkun, Bing Xiang, Matt Gardner, Sameer Singh

Question Generation Based on Grammar Knowledge and Fine-grained Classification (2022.coling-1)

Challenge:	Recent studies have shown that discriminative training results in models that exploit these underlying biases to achieve a better held-out performance, without learning the right way to reason.
Approach:	They propose a generative context selection model for multi-hop QA that reasons about how the given question could have been generated given a context pair and not just independent contexts.
Outcome:	The proposed model outperforms the state-of-the-art model on hotpotQA while being comparable to the state of the art answering performance on adversarial held-out set.

Copied to clipboard

Yuan Sun, Sisi Liu, Zhengcuo Dan, Xiaobing Zhao

Information Extraction & Text Mining

Challenge:	Recent research on question generation has achieved great success, but some question types and answers did not match.
Approach:	They construct a question type classifier and a query generator to solve the problem of question types not matching with other questions.
Outcome:	The proposed model improves the accuracy of interrogative words in generated questions.

Is Graph Structure Necessary for Multi-hop Question Answering? (2020.emnlp-main)

Copied to clipboard

Nan Shao, Yiming Cui, Ting Liu, Shijin Wang, Guoping Hu

Structured Data in NLP

Challenge:	Existing studies focus on multi-hop question answering across multiple documents or paragraphs.
Approach:	They propose a graph neural network to deal with graph structure in textual multi-hop reasoning . they propose 'self-attention' and propose removing entire graph structure may not hurt the final results .
Outcome:	The proposed model shows that graph-attention or the entire graph structure can be replaced by self-attention . hotpotQA is a widely used benchmark for multi-hop question answering .

Multi-hop Reading Comprehension through Question Decomposition and Rescoring (P19-1)

Copied to clipboard

Sewon Min, Victor Zhong, Luke Zettlemoyer, Hannaneh Hajishirzi

Machine Reading Comprehension

Challenge:	Existing systems for multi-hop reading comprehension decompose compositional questions into simpler sub-questions . authors propose a system that learns to break compositional multi- hop questions into simple singlehop sub-question .
Approach:	They propose a system that decomposes a compositional question into simpler sub-questions . they propose recast subquestion generation as a span prediction problem .
Outcome:	The proposed system generates as effective as human-authored sub-questions using 400 examples . it also provides explainable evidence for its decision making in the form of sub-questions .

Dynamically Fused Graph Network for Multi-hop Reasoning (P19-1)

Copied to clipboard

Lin Qiu, Yunxuan Xiao, Yanru Qu, Hao Zhou, Lei Li, Weinan Zhang, Yong Yu

Multimodality

Challenge:	Text-based question answering (TBQA) has been studied extensively in recent years.
Approach:	They propose a Dynamically Fused Graph Network to answer questions requiring multiple scattered evidence and reasoning over them.
Outcome:	The proposed method achieves competitive results on a public TBQA dataset and produces interpretable reasoning chains.

CoAT: Chain-of-Associated-Thoughts Framework for Enhancing Large Language Models Reasoning (2025.findings-emnlp)

Copied to clipboard

Jianfeng Pan, Senyou Deng, Shaomang Huang

Graph-Induced Transformers for Efficient Multi-Hop Question Answering (2022.emnlp-main)

Challenge:	OpenAI-o1 enables ‘slow thinking’ because it is closer to the human thought process .
Approach:	They propose a new framework that integrates the Monte Carlo Tree Search algorithm and a dynamic mechanism for integrating new key information, termed ‘associative memory’.
Outcome:	The proposed framework improves performance on open-source multi-hop reasoning datasets and more than 15% gain on proprietary CRB dataset.

Copied to clipboard

Giwon Hong, Jeonghwan Kim, Junmo Kang, Sung-Hyon Myaeng

A Simple Yet Strong Pipeline for HotpotQA (2020.emnlp-main)

Challenge:	Recent MHQA tasks that require inter-paragraph/sentence linkages use graphs to model internal structural information within text.
Approach:	They propose a graph-induced transformer that applies graph-derived attention patterns directly into a PLM without external graph modules.
Outcome:	The proposed model can replace external graph modules while preserving model performance.

Copied to clipboard

Dirk Groeneveld, Tushar Khot, null Mausam, Ashish Sabharwal

Is Multihop QA in DiRe Condition? Measuring and Reducing Disconnected Reasoning (2020.emnlp-main)

Challenge:	Existing models for multi-hop question answering have been proposed with varying complexities.
Approach:	They propose to use BERT to identify potentially relevant sentences independently of each other . they feed selected sentences into a standard BERT span prediction model to choose an answer .
Outcome:	The proposed pipeline outperforms existing models on hotpotQA and support identification.

Copied to clipboard

Harsh Trivedi, Niranjan Balasubramanian, Tushar Khot, Ashish Sabharwal

Unsupervised Question Decomposition for Question Answering (2020.emnlp-main)

Challenge:	Existing models exploit dataset artifacts to produce correct answers without connecting information across multiple facts.
Approach:	They formalize disconnected reasoning across subsets of supporting facts to reduce disconnected reasoning . they propose an automatic transformation of existing datasets that reduces disconnected reasoning.
Outcome:	The proposed model-agnostic probe reduces disconnected reasoning in a reading comprehension setting.

Copied to clipboard

Ethan Perez, Patrick Lewis, Wen-tau Yih, Kyunghyun Cho, Douwe Kiela

Low-Resource NLP

Challenge:	Existing QA systems struggle to answer complex questions because information is scattered in different places.
Approach:	They propose an unsupervised algorithm that decomposes hard questions into simpler sub-questions . they propose an algorithm that can be used to generate a final answer from millions of questions .
Outcome:	The proposed algorithm decomposes hard questions into simpler sub-questions that existing QA systems can answer.

HOLMES: Hyper-Relational Knowledge Graphs for Multi-hop Question Answering using LLMs (2024.acl-long)

Copied to clipboard

Pranoy Panda, Ankush Agarwal, Chaitanya Devaguptapu, Manohar Kaul, Prathosh Ap

Reference-based Metrics Disprove Themselves in Question Generation (2024.findings-emnlp)

Challenge:	Existing approaches to answer multi-hop questions are query-agnostic and the extracted facts are ambiguous as they lack context.
Approach:	They propose to use a knowledge graph to extract query-relevant information from unstructured text.
Outcome:	The proposed method achieves performance improvements on two popular datasets.

Copied to clipboard

Bang Nguyen, Mengxia Yu, Yun Huang, Meng Jiang

Text Generation

Challenge:	Existing metrics for question generation are based on human-written references . however, the results of the metrics on our study disprove the metrics themselves .
Approach:	They propose to use human-written references to evaluate question generation . they propose to combine criteria such as naturalness, answerability, and complexity .
Outcome:	The proposed model is based on multi-dimensional criteria such as naturalness, answerability, and complexity, utilizing large language models.

AnchorCoT: Anchors Pave the Way for Multi-hop Reasoning (2025.findings-acl)

Copied to clipboard

Tianshi Ming, Xian Wu, Yingying Zhang, Zichuan Fu, Dawei Cheng

LCR-RAG: Enhancing Logical Consistency in Retrieval-Augmented Generation via Neuro-symbolic Reinforcement Learning (2026.acl-long)

Challenge:	Large language models (LLMs) have demonstrated potential reasoning capabilities through prompt design, such as the Chain of Thought (CoT).
Approach:	They propose a new reasoning approach that predicts key entities which work as important “anchors” and employs a ranking algorithm to ensure the logical sequence of the predicted answers.
Outcome:	The proposed approach outperforms existing methods in multi-hop question reasoning and provides more accurate reasoning results in multihop question answering tasks.

Copied to clipboard

Wenxiang Zheng, Guo Tang, Shixin Jiang, Liangyu Huo, Xiyuan Zhang, Jian Xie, Ming Liu

Optimizing Multi-Hop Document Retrieval Through Intermediate Representations (2025.findings-acl)

Challenge:	Retrieval-Augmented Generation (RAG) is widely used to ground large language models in external knowledge and improve factual accuracy.
Approach:	They propose a framework that integrates neuro-symbolic verification with reinforcement learning to optimize logical consistency.
Outcome:	The proposed framework outperforms strong RAG baselines on hotpotQA, ASQA, and TriviaQA.

Copied to clipboard

Linjiaen Linjiaen, Jingyu Liu, Yingbo Liu

Re-ReST: Reflection-Reinforced Self-Training for Language Agents (2024.emnlp-main)

Challenge:	Existing approaches to addressing multi-hop queries are computationally expensive . despite their success, large language models often generate factually incorrect answers .
Approach:	They propose a layer-by-layer reasoning approach that leverages intermediate representations from the middle layers to retrieve external knowledge.
Outcome:	The proposed method outperforms existing RAG methods on open-domain multi-hop question-answering datasets while maintaining inference overhead similar to that of standard RAG.

Copied to clipboard

Zi-Yi Dou, Cheng-Fu Yang, Xueqing Wu, Kai-Wei Chang, Nanyun Peng

Few-shot Reranking for Multi-hop QA via Language Model Prompting (2023.acl-long)

Challenge:	Existing methods to fine tune language agents with reasoning-action trajectories require high-quality model-generated samples, which are hard to obtain for challenging language agent tasks.
Approach:	They propose a method to employ reflection during inference without ground-truth feedback to improve agents more autonomously.
Outcome:	The proposed method improves self-training performance on open-source language agents by 7.6% and 14.1% respectively.

Copied to clipboard

Muhammad Khalifa, Lajanugen Logeswaran, Moontae Lee, Honglak Lee, Lu Wang

DAVIS: Planning Agent with Knowledge Graph-Powered Inner Monologue (2025.findings-emnlp)

Challenge:	Existing methods for multi-hop QA with open-domain questions require a large number of labeled question-document pairs for retrieval.
Approach:	They propose a language-based prompt for multi-hop path reranking that relies on language model prompting to generate a relevance score between a question and the path.
Outcome:	The proposed method yields strong retrieval performance on HotpotQA with only 128 training examples compared to state-of-the-art methods trained on thousands of examples.

Copied to clipboard

Minh Pham Dinh, Michael G Yankoski, Munira Syed, Trenton W. Ford

Knowledge Representation

Challenge:	Existing approaches to designing a generalist scientific agent fail to address multifaceted requirements of scientific tasks.
Approach:	DAVIS is a generalist scientific agent capable of performing tasks in laboratory settings to assist researchers.
Outcome:	DAVIS performs better on ScienceWorld benchmarks compared to previous approaches on 8 out of 9 elementary science subjects.

ChatCoT: Tool-Augmented Chain-of-Thought Reasoning on Chat-based Large Language Models (2023.findings-emnlp)

Copied to clipboard

Zhipeng Chen, Kun Zhou, Beichen Zhang, Zheng Gong, Xin Zhao, Ji-Rong Wen

LastingBench: Defend Benchmarks Against Knowledge Leakage (2025.findings-emnlp)

Challenge:	Large language models (LLMs) have excellent performance in evaluation benchmarks, but struggle in complex reasoning tasks.
Approach:	They propose a tool-augmented chain-of-thought reasoning framework for chat-based LLMs . they model chain- of-thoughting reasoning as multi-turn conversations to utilize tools .
Outcome:	The proposed framework can outperform state-of-the-art models on complex reasoning tasks.

Copied to clipboard

Yixiong Fang, Tianran Sun, Yuling Shi, Min Wang, Xiaodong Gu

Mitigating Lost-in-Retrieval Problems in Retrieval Augmented Multi-Hop Question Answering (2025.acl-long)

Challenge:	Existing methods to detect and safeguard LLMs against knowledge leakage fail to address the long-term challenge of mitigating it.
Approach:	They propose a method to reinforce and safeguard existing benchmarks against knowledge leakage by perturbation-based detection and counterfactual rewriting to disrupt memorization while preserving original intent.
Outcome:	The proposed method reduces memorization effects in long-context QA benchmarks, providing a more accurate assessment of model reasoning and generalization abilities.

Copied to clipboard

Rongzhi Zhu, Xiangyu Liu, Zequn Sun, Yiwei Wang, Wei Hu

Prompting Explicit and Implicit Knowledge for Multi-hop Question Answering Based on Human Reading Process (2024.lrec-main)

Challenge:	Empirical results show that ChainRAG consistently outperforms baselines in both effectiveness and efficiency.
Approach:	They propose a method which sequentially handles each sub-question by completing missing key entities and retrieving relevant sentences from a sentence graph for answer generation.
Outcome:	The proposed method outperforms baselines on three multi-hop QA datasets.

Copied to clipboard

Guangming Huang, Yunfei Long, Cunjin Luo, Jiaxing Shen, Xia Sun

KCS: Diversify Multi-hop Question Generation with Knowledge Composition Sampling (2025.emnlp-main)

Challenge:	Existing studies have not explored the link between PLMs’ pre-training-based knowledge and input passages.
Approach:	They propose a framework that uses prompts to connect explicit and implicit knowledge to elicit type-specific reasoning via prompts, a form of implicit knowledge.
Outcome:	The proposed model performs comparable to the state-of-the-art on HotpotQA.

Copied to clipboard

Yangfan Wang, Jie Liu, Chen Tang, Lian Yan, Jingchi Jiang

Text Generation

Challenge:	Existing approaches to multi-hop question answering focus on generating simple questions and neglecting the integration of essential knowledge, such as relevant sentences within documents.
Approach:	They propose a framework to expand the diversity of generated multi-hop questions by sampling varied knowledge compositions within a given context.
Outcome:	The proposed framework improves the overall accuracy of knowledge composition selection by 3.9% on hotpotQA and 2WikiMultihopQA datasets.

Towards Fully Exploiting LLM Internal States to Enhance Knowledge Boundary Perception (2025.acl-long)

Copied to clipboard

Shiyu Ni, Keping Bi, Jiafeng Guo, Lulu Yu, Baolong Bi, Xueqi Cheng

S2G-RAG: Structured Sufficiency and Gap Judging for Iterative Retrieval-Augmented QA (2026.acl-long)

Challenge:	Large language models (LLMs) exhibit impressive performance across diverse tasks but struggle to accurately gauge their knowledge boundaries.
Approach:	They propose Consistency-based Confidence Calibration (C3) which assesses confidence consistency through question reformulation to improve LLMs’ ability to recognize their knowledge gaps.
Outcome:	The proposed method improves the unknown perception rate by 5.6% on NQ and 4.9% on HotpotQA.

Copied to clipboard

Minghan Li, Junjie Zou, Xinxuan Lv, Chao Zhang, Guodong Zhou

Concept Tokens: Learning Behavioral Embeddings Through Concept Definitions (2026.findings-acl)

Challenge:	Retrieval-augmented generation grounds language models in external evidence, but multi-hop question answering remains difficult . iterative pipelines must control what to retrieve next and when evidence is adequate.
Approach:	They propose an iterative framework with an explicit controller, S2G-Judge . they map structured gap items into the next retrieval query to produce stable retrieval trajectories .
Outcome:	Experiments on TriviaQA, HotpotQA, and 2WikiMultiHopQA show that S2G-RAG improves multi-hop QA performance and robustness under multi-turn retrieval.

Copied to clipboard

Ignacio Sastre, Aiala Rosá

Representation Learning

Challenge:	Concept Tokens is a lightweight method that adds a special token to a pretrained LLM . we find that negating the hallucination token reduces hallucines and lowers precision .
Approach:	They propose a lightweight method that adds a new special token to a pretrained LLM and learns only its embedding from multiple natural language definitions of a target concept.
Outcome:	The proposed method can learn only its embedding from multiple definitions of a target concept . the study shows that it can improve hallucinations and recasting in closed-book questions .

Lexical Diversity-aware Relevance Assessment for Retrieval-Augmented Generation (2025.acl-long)

Copied to clipboard

Zhange Zhang, Yuqing Ma, Yulong Wang, Shan He, Tianbo Wang, Siqi He, Jiakai Wang, Xianglong Liu

Search-in-Context: Efficient Multi-Hop QA over Long Contexts via Monte Carlo Tree Search with Dynamic KV Retrieval (2025.findings-acl)

Challenge:	Extensive experiments on widely used benchmarks demonstrate the efficacy of our approach, yielding a 10.6% accuracy improvement on HotpotQA.
Approach:	They propose a Lexical Diversity-aware RAG method to address the biases in relevant information retrieval and utilization induced by lexical diversity.
Outcome:	Extensive experiments on widely used benchmarks show the proposed method yields a 10.6% accuracy improvement on HotpotQA.

Copied to clipboard

Jiabei Chen, Guang Liu, Shizhu He, Kun Luo, Yao Xu, Jun Zhao, Kang Liu

IterCOMP: Reasoning-aware Adaptive Prompt Compression for Multi-hop Question Answering (2026.acl-long)

Challenge:	Existing approaches to multihop question answering (MHQA) over long contexts are often neglecting explicit reasoning or incurring expensive computational costs due to full-attention mechanisms over long contextuals.
Approach:	They propose a framework that integrates Monte Carlo Tree Search (MCTS) with dynamic key-value retrieval to enable iterative, context-aware reasoning.
Outcome:	The proposed framework integrates Monte Carlo Tree Search (MCTS) with dynamic key-value (KV) retrieval to enable iterative, context-aware reasoning.

Copied to clipboard

JungMin Yun, YoungBin Kim

Low-Resource NLP

Challenge:	Existing prompt compression methods are designed for single-turn queries and fail to capture interdependent reasoning steps.
Approach:	They propose a unified, training-free prompt compression framework that integrates multi-hop reasoning within an iterative compression loop.
Outcome:	Experiments on MusiQue, 2WikiMultiHopQA, and HotpotQA show that iterCOMP achieves significant improvements in Exact Match and F1 scores while reducing the token budget.

RouterHGC: Optimized Router for LLM-based Multi-Agent Systems via Heterogeneous Graph Contrastive Learning (2026.findings-acl)

Copied to clipboard

Yitao Xiao, Shaoyong Guo, Guoming Yang, Qingnan Wang, Yinlin Ren, Xuesong Qiu, Qi Feng