Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

Hierarchical Lexical Graph for Enhanced Multi-Hop Retrieval

T0 review · 5 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that indexing documents as atomic propositions linked by shared entities in a three-tier graph lets retrieval-augmented generation answer multi-hop questions more accurately than chunk-based retrieval, with an average…

desk verdict Solid system, shaky mechanism: gains over chunk RAG are credible but the graph's contribution needs a random-link ablation before trusting the headline numbers. read the letter →

arxiv 2506.08074 v1 pith:YINYNYWO submitted 2025-06-09 cs.IR cs.AIcs.CL

Abdellah Ghassel , Ian Robinson , Gabriel Tanase , Hal Cooper , Bryan Thompson , Zhen Han , Vassilis N. Ioannidis , Soji Adeshina
show 1 more author
Huzefa Rangwala
This is my paper
classification cs.IRcs.AIcs.CL
keywords multi-hopretrievalretrieval-augmentedgenerationhierarchicallexicalgraphpropositiongranularitybeamsearchentitylinkingsyntheticquestionansweringdataset
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that the failure of retrieval-augmented generation on multi-hop questions comes from retrieving oversized text chunks by vector similarity alone, and that the fix is a three-tier index called the Hierarchical Lexical Graph (HLG). HLG stores atomic propositions as the retrieval unit, groups them into topics, and links them through shared entities and relations so a retriever can walk across documents. On top of it the paper builds two retrievers: StatementGraphRAG, which runs an entity-aware beam search over propositions, and TopicGraphRAG, which selects topics and then expands along entity links. Across five datasets the methods beat chunk-based RAG baselines, with an average relative improvement of 23.1% in retrieval recall and correctness, and a synthetic dataset pipeline is introduced to create harder multi-document questions. If true, this means multi-hop QA can be improved without changing the generator, only by indexing evidence at finer granularity and making the index traversable.

What carries the argument

The load-bearing object is the Hierarchical Lexical Graph (HLG), an index with three tiers: source lineage, topic clusters, and an entity-relationship layer. The retrieval mechanism that carries the argument is graph beam search (Algorithm 1). Starting from keyword-matched and vector-similar statements, it expands each statement to neighbours that share at least one entity, $Nbr(s)=\{s'\in \mathcal{S}_G \mid \mathrm{Ent}(s)\cap\mathrm{Ent}(s')\neq\emptyset\}$, scores each path by the cosine similarity between the query embedding and an attention-weighted sum of the statement embeddings along the path, $Score_{\mathrm{beam}}(P)=\mathrm{sim}(\mathbf{e}_Q,\sum_i \alpha_i \mathbf{e}_{s_i})$, and then reranks the expanded pool with a cross-encoder reranker. The central hypothesis is that shared entities are the bridge between semantically distant statements that vector search alone would miss.

What would settle it

A direct test would be to run StatementGraphRAG and TopicGraphRAG on a multi-hop dataset with the graph beam-search step replaced by random expansion to the same number of neighbours, holding retrieval and reranking fixed; if random expansions match the entity-expansion results, shared-entity edges are not carrying the gain. A second check is to measure retrieval recall as a function of the number of true hops and compare passages reached only through entity edges against passages reached by vector similarity alone.

Watch

Extended reading notes

Core claim

The paper sets out to establish that the retrieval bottleneck in multi-hop question answering is not the generator but the granularity and connectivity of the index. It builds the Hierarchical Lexical Graph (HLG), a three-tier structure in which every atomic proposition is traced to its source chunk (Lineage Tier), grouped into thematic topics (Summarization Tier), and linked to other propositions through entities and relations (Entity-Relationship Tier). On top of HLG, StatementGraphRAG runs an entity-aware beam search over propositions to answer precise factoid questions, while TopicGraphRAG selects topics first and then expands along entity links for broader exploratory questions. The paper reports that across MultiHop-RAG, SEC-10Q, ConcurrentQA, NTSB, and WikiHowQA these methods outperform chunk-based RAG baselines, with an average relative improvement of 23.1% in retrieval recall and correctness, and it introduces a synthetic pipeline that generates 674 validated multi-document question-answer pairs from the MultiHop-RAG corpus.

Load-bearing premise

The method's multi-hop advantage rests on the premise that two statements sharing an entity are likely to be steps in one evidence chain, so expanding along entity-overlap edges adds relevant evidence rather than noise; the paper does not separately measure how often entity extraction or linking mistakes break that premise.

Editorial extensions

If this is right

  • Statement-level retrieval with graph expansion raises correctness on single-answer multi-hop datasets, with the highest average correctness of 73.6% for SGRAG-0.5% versus 66.1% for the reranked chunk baseline.
  • Topic-level retrieval with graph expansion raises answer recall on multi-answer datasets, with TGRAG reaching a 53.8% average recall.
  • The gains survive chunk-constrained generation: Chunk-SGRAG and Chunk-TGRAG outperform chunk-only baselines, so graph expansion helps even when the final prompt must use original text blocks.
  • Graph expansion can hurt single-hop queries: on WikiHowQA the simpler chunk baseline wins, so multi-hop traversal should be paired with early stopping for single-hop questions.
  • The synthetic pipeline yields 674 validated multi-hop questions from the MultiHop-RAG corpus and can serve as harder evaluation data for multi-hop retrieval systems.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Beyond the paper, the biggest practical risk is that the graph's value depends on entity extraction quality; a deployment should measure entity precision and recall first, since noisy shared-entity edges would add irrelevant statements rather than bridge documents.
  • The authors report indexing the MultiHop-RAG corpus in under an hour at about $145 of LLM cost, which makes HLG attractive for corpora queried many times but expensive for one-shot analytical workloads.
  • A natural next experiment the paper leaves implicit is comparing shared-entity edges with typed or temporal relation edges; typed edges might reduce the over-expansion error the authors observe on highly connected entities.
  • Because the synthetic query pipeline filters questions through an internal retriever, the resulting benchmark may favour systems similar to that retriever; an external human audit of rejected questions would test for that bias.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper introduces the Hierarchical Lexical Graph (HLG), a three-tier index over atomic propositions, topics, and entity-relationship triples, and two retrievers built on it: StatementGraphRAG, which combines keyword and vector search with beam search over entity-sharing statements, and TopicGraphRAG, which retrieves topics and expands along entity links. The paper also contributes a synthetic multi-hop QA generation pipeline based on the MultiHop-RAG corpus, yielding 674 validated question-answer pairs. Experiments on MultiHop-RAG, SEC-10Q, ConcurrentQA, NTSB, and WikiHowQA compare HLG variants against chunk-based RAG baselines on correctness, answer recall, RAGChecker metrics, and pairwise LLM win rates. The reported results show consistent gains for HLG variants, with SGRAG-0.5% achieving the highest average correctness (73.6%) and TGRAG the highest average answer recall (53.8%).

Significance. If the graph-traversal mechanism is what drives the reported gains, this is a useful contribution: fine-grained proposition retrieval combined with cross-document entity paths is a plausible and practical direction for multi-hop RAG, and the synthetic benchmark pipeline addresses a real gap in existing evaluation sets. The paper has several strengths: it evaluates on five independent public datasets, includes an entity-linking baseline (E1), reports statement-fidelity validation, provides indexing cost and latency details, and links to an open-source library. However, the paper does not isolate the graph-traversal mechanism from candidate-pool effects, and several evaluation choices are under-specified. The central mechanism should be validated through controlled ablations before the performance claims can be fully accepted.

major comments (5)
  1. [§4.1, §6.3.2, Table 2] The only direct comparison isolating the graph-beam-search mechanism is SGRAG vs. SRAG: SGRAG improves average correctness from 69.2% to 73.6% and average recall from 49.2% to 52.4%. But SGRAG starts from SRAG's 100 statements and adds beam-search neighbours before the same reranker, so the gain may be a candidate-pool-size effect rather than evidence that entity-overlap traversal finds correct multi-hop bridges. Add a controlled ablation in which SRAG is given a comparable number of additional statements (e.g., 200 or 300 via repeated VSS, or random neighbours) and report whether entity-overlap expansion still improves over the same-size pool. Without this, the central claim that graph traversal is beneficial is not established.
  2. [§3.2/Eq. (5), Appendix A, §C.2.1] The graph mechanism relies entirely on Ent(s)∩Ent(s′) edges, but no evaluation of entity extraction or linking accuracy is reported. Appendix A validates statement fidelity, not entity precision/recall, and §C.2.1 concedes that supernodes cause over-expansion. Please report entity extraction/linking quality (or edge-relevance judgments) and include a random-edge or degree-matched ablation. If entity edges are noisy, the expansion may simply be adding larger pools of loosely relevant statements, which is structurally different from 'exposing cross-document paths'.
  3. [§6.5, Tables 2–5] The evaluation protocol is under-specified for a paper whose claims are entirely empirical. The LLM judge used for correctness/answer recall is not identified (model, prompt, temperature, or agreement with human judges), and no confidence intervals or significance tests are reported; several headline differences are small (e.g., SGRAG-0.5% 73.6% vs. TGRAG 72.2% correctness; TGRAG 53.8% vs. SGRAG-0.5% 52.4% recall). Please provide the judge details and at least paired bootstrap or significance tests across queries.
  4. [§6.3, §6.4] The token budget for statement-level methods is not specified. §6.3 says the context window is fixed at 10 chunks (~3,000 tokens), but SGRAG retrieves 100 statements and the final number of statements returned is never stated; if statement methods effectively receive more tokens than chunk baselines, the comparison is unfair. Specify the final context budget (number of statements or tokens) for every method and dataset.
  5. [§5.2, §7.3, Table 4] The synthetic dataset is generated using HLG as its backbone (topic collection and chunk selection in §5.2) and validated with an 'internal retriever pipeline', so Table 4's RAGChecker comparison is at risk of circularity: HLG-based retrievers may be advantaged because the benchmark was constructed from their own topic/entity structures. The independent datasets in Tables 2–3 are more persuasive; please either show that the synthetic pipeline does not depend on HLG-specific structures or present the synthetic evaluation as an auxiliary sanity check rather than as evidence for the main claim.
minor comments (5)
  1. [Abstract, §7] The abstract's 'average relative improvement of 23.1%' is not reproducible from Table 2; specify which methods, which baseline, and which definition of relative improvement produce this number.
  2. [§4.2] TopicGraphRAG is described only qualitatively; it needs equations or pseudocode comparable to Algorithm 1 so that the reported results are reproducible.
  3. [§6.3.2] Beam width B=50, depth D_max=3, and the diversity threshold τ=0.5% are fixed without a sensitivity analysis; at least one variation of B and D_max would help establish robustness.
  4. [Table 2] The bold/underline convention is ambiguous when values tie at 86.9% in the MultiHop-RAG correctness column; clarify how ties are treated.
  5. [§6.2] The indexing procedure samples 'five chunks' for domain-adaptive refinement; it would be useful to state how representative this sample is for each corpus and whether results are stable across samples.

Circularity Check

1 steps flagged · score 3.0 of 10

Synthetic benchmark circularity is localized; the core external-benchmark claim remains independent.

  1. self definitional [Section 1 (Introduction), Section 5.2 (Pipeline Architecture), Section 7.3 (Synthetic Dataset Evaluation)]
    "we present a synthetic multi-hop summarization pipeline using HLG as a backbone. ... (1) Topic Collection. From a seed topic, we retrieve semantically related topics from different documents. (2) Chunk Selection. Collect chunks from each relevant topic (3-5 distinct articles). ... An internal retriever pipeline is used to simulate the reasoning path. A query is accepted only if the system can reconstruct the ground truth from the provided snippets. ... We further investigated retrieval quality on our synthetic MultiHop-RAG subset using RAGChecker [19] ..."

    The synthetic benchmark is generated by the very graph being evaluated. The paper states the pipeline uses HLG as a backbone; topic collection retrieves semantically related topics from different documents through HLG's topic/entity structure, and chunk selection takes chunks from those HLG-chosen topics. The acceptance filter then employs an internal retriever pipeline that accepts only queries whose evidence can be reconstructed from the provided snippets. Thus the ground-truth multi-hop relevance in this subset is partly defined by HLG's own connections and by its own retriever behavior. Reporting SGRAG/TGRAG gains on this self-constructed subset (Table 4, Table 5) is therefore partially circular: the task rewards the same graph relations used to create it.

full rationale

The core retrieval comparison on MultiHop-RAG, SEC-10Q, ConcurrentQA, NTSB, and WikiHowQA is externally grounded: the graph structures are built from the corpora, and the baselines are standard chunk-based VSS with and without reranking. No equation in Section 4 reduces the measured gains to a fitted parameter or to a self-citation; the graph beam search is a real algorithmic transformation of the candidate pool. The only circular element found is the synthetic dataset: because HLG is used as the backbone to select related topics and chunks, and an internal retriever pipeline filters queries, Table 4's evidence for TGRAG/SGRAG is partly self-confirming. Since the headline average improvement does not depend only on this synthetic evaluation, the overall circularity is limited rather than pervasive.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The paper rests on two categories of assumptions: the faithfulness of LLM-based graph construction (propositions, entities, topics) and the validity of the scoring and evaluation heuristics. These are reasonable for an applied systems paper, but several are only weakly validated, especially entity linking and LLM-as-judge correctness. No new physical or formal entities are introduced.

free parameters (6)
  • Beam width B = 50
    Chosen for graph beam search in SGRAG and TGRAG (Section 6.3.2); no justification is given for this value.
  • Maximum depth D_max = 3
    Graph expansion depth in beam search (Section 6.3.2); set by hand, not analyzed.
  • Diversity threshold tau = 0.5%
    Empirically selected on the evaluation datasets to improve correctness (Section 4.3.1, Table 2); this is post-hoc tuning on the test set.
  • Initial retrieval count k = 100 statements / 50 topics
    Number of candidates retrieved before graph expansion and reranking (Section 6.3.2).
  • Context window size = 10 chunks (~3000 tokens)
    Fixed generation context for all methods (Section 6.3).
  • Chunk size = 300 tokens, 20% overlap
    Chunking configuration used for all datasets (Section 6.2).
assumptions (5)
  • domain assumption LLM-based proposition extraction preserves the semantic content of source chunks.
    The entire graph is built on propositions extracted by Claude-3 Sonnet; if these are lossy, retrieval quality degrades. Validated only via a sampled accuracy check (Appendix A).
  • domain assumption Shared-entity edges in the Entity-Relationship tier connect evidence relevant to multi-hop queries.
    Beam search over entity neighbors (Eq. 5) assumes that entity overlap is a reliable signal for cross-document relevance; no entity-linking quality evaluation is provided.
  • domain assumption Attention-weighted path embedding (Eq. 6) with cosine similarity is an effective path scoring function.
    The path relevance score (Eq. 7) is a heuristic; no comparison to alternative scoring methods is given.
  • domain assumption LLM-based correctness judgments align with human judgments.
    The primary metric relies on an unspecified LLM evaluator; the paper does not report human agreement or the evaluator's identity.
  • domain assumption Synthetic dataset queries generated by the pipeline genuinely require multi-hop reasoning.
    Validation is performed by the same LLM pipeline (Section 5.2); no independent human assessment is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hierarchical Lexical Graph for Enhanced Multi-Hop Retrieval." pith.science (2026). https://pith.science/paper/YINYNYWO

@misc{pith2026250608074,
  author       = {Pith},
  title        = {Pith review of: Hierarchical Lexical Graph for Enhanced Multi-Hop Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YINYNYWO}},
  note         = {Machine review of arXiv:2506.08074}
}
read the original abstract

Retrieval-Augmented Generation (RAG) grounds large language models in external evidence, yet it still falters when answers must be pieced together across semantically distant documents. We close this gap with the Hierarchical Lexical Graph (HLG), a three-tier index that (i) traces every atomic proposition to its source, (ii) clusters propositions into latent topics, and (iii) links entities and relations to expose cross-document paths. On top of HLG we build two complementary, plug-and-play retrievers: StatementGraphRAG, which performs fine-grained entity-aware beam search over propositions for high-precision factoid questions, and TopicGraphRAG, which selects coarse topics before expanding along entity links to supply broad yet relevant context for exploratory queries. Additionally, existing benchmarks lack the complexity required to rigorously evaluate multi-hop summarization systems, often focusing on single-document queries or limited datasets. To address this, we introduce a synthetic dataset generation pipeline that curates realistic, multi-document question-answer pairs, enabling robust evaluation of multi-hop retrieval systems. Extensive experiments across five datasets demonstrate that our methods outperform naive chunk-based RAG achieving an average relative improvement of 23.1% in retrieval recall and correctness. Open-source Python library is available at https://github.com/awslabs/graphrag-toolkit.

Figures

Figures reproduced from arXiv: 2506.08074 by the authors.

Figure 1
Figure 1. Hierarchical Lexical Graph Model. Statements are [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. KAMR: Grounding Generation via Knowledge-Aligned Multi-hop Retrieval

    cs.IR 2026-07 conditional novelty 6.0 of 10

    Partial-alignment contrastive pretraining plus anchor-then-expand graph retrieval improves multi-hop KG evidence recovery and downstream QA over strong dense and graph RAG baselines.

Reference graph

Works this paper leans on

24 extracted references · 12 canonical work pages · cited by 1 Pith paper

  1. [1]

    Anthropic. 2024. Claude 3 Model Card. https://www.anthropic.com/model_ cards/claude_3.pdf

    work page 2024

  2. [2]

    Anthropic. 2024. Model Card Addendum: Claude 3.5 Haiku and Upgraded Claude 3.5 Sonnet. https://assets.anthropic.com/m/1cd9d098ac3e6467/original/Claude- 3-Model-Card-October-Addendum.pdf

    work page 2024

  3. [3]

    Simran Arora, Patrick Lewis, Angela Fan, Jacob Kahn, and Christopher Ré. 2023. Reasoning over Public and Private Data in Retrieval-Based Systems.Transactions of the Association for Computational Linguistics11 (2023), 902–921. https://doi. org/10.1162/tacl_a_00580

    work page doi:10.1162/tacl_a_00580 2023

  4. [4]

    Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2016. Neural Machine Translation by Jointly Learning to Align and Translate. arXiv:1409.0473 [cs.CL] https://arxiv.org/abs/1409.0473

    arXiv 2016

  5. [5]

    Maciej Besta, Ales Kubicek, Roman Niggli, Robert Gerstenberger, Lucas Weitzen- dorf, Mingyuan Chi, Patrick Iff, Joanna Gajda, Piotr Nyczyk, Jürgen Müller, Hubert Niewiadomski, Marcin Chrapek, Michał Podstawski, and Torsten Hoe- fler. 2024. Multi-Head RAG: Solving Multi-Aspect Problems with LLMs. arXiv:2406.05085 [cs.CL] https://arxiv.org/abs/2406.05085

    arXiv 2024

  6. [6]

    Valeriia Bolotova-Baranova, Vladislav Blinov, Sofya Filippova, Falk Scholer, and Mark Sanderson. 2023. WikiHowQA: A Comprehensive Benchmark for Multi- Document Non-Factoid Question Answering. InProceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Anna Rogers, Jordan Boyd-Graber, and Naoaki Okazak...

    work page 2023

  7. [7]

    Tong Chen, Hongwei Wang, Sihao Chen, Wenhao Yu, Kaixin Ma, Xinran Zhao, Hongming Zhang, and Dong Yu. 2024. Dense X Retrieval: What Retrieval Granu- larity Should We Use?. InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, Yaser Al-Onaizan, Mohit Bansal, and Yun-Nung Chen (Eds.). Association for Computational Linguist...

    doi:10.18653/v1/2024.emnlp-main.845 2024

  8. [8]

    Cohere. 2023. Introducing Embed v3. https://cohere.com/blog/introducing- embed-v3. Accessed: 2025-05-27

    work page 2023

Show all 24 references
  1. [9]

    Cover and P

    T. Cover and P. Hart. 1967. Nearest neighbor pattern classification.IEEE Trans- actions on Information Theory13, 1 (1967), 21–27. https://doi.org/10.1109/TIT. 1967.1053964

    1967

  2. [10]

    Nicola De Cao, Wilker Aziz, and Ivan Titov. 2019. Question Answering by Reasoning Across Documents with Graph Convolutional Networks. InProceed- ings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technolog...

    2019 doi

  3. [11]

    Darren Edge, Ha Trinh, Newman Cheng, Joshua Bradley, Alex Chao, Apurva Mody, Steven Truitt, and Jonathan Larson. 2024. From Local to Global: A Graph RAG Approach to Query-Focused Summarization. arXiv:2404.16130 [cs.CL] https://arxiv.org/abs/2404.16130

    2024 arXiv

  4. [12]

    Yuwei Fang, Siqi Sun, Zhe Gan, Rohit Pillai, Shuohang Wang, and Jingjing Liu. 2020. Hierarchical Graph Network for Multi-hop Question Answering. InProceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Bonnie Webber, Trevor Cohn, Yulan ...

    2020 doi

  5. [13]

    Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, Qianyu Guo, Meng Wang, and Haofen Wang. 2023. Retrieval-Augmented Generation for Large Language Models: A Survey. CoRRabs/2312.10997 (2023). https://doi.org/10.48550/ARXIV.2312.10997 arXi...

    2023 doi arXiv

  6. [14]

    Yunjie He, Philip John Gorinski, Ieva Staliunaite, and Pontus Stenetorp

  7. [15]

    Taqi Jaffri. 2023. Announcing Docugami Knowledge Graph Retrieval Augmented Generation (KG-RAG) Datasets in the LlamaHub. https://www.docugami.com/ blog/kg-rag-datasets-llama-index

    2023

  8. [16]

    Seungone Kim, Juyoung Suk, Shayne Longpre, Bill Yuchen Lin, Jamin Shin, Sean Welleck, Graham Neubig, Moontae Lee, Kyungjae Lee, and Minjoon Seo. 2024. Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Models. InProceedings of the 2024 Confere...

    2024 doi

  9. [17]

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive NLP tasks. InProceedings o...

    2020

  10. [18]

    Chaofan Li, Zheng Liu, Shitao Xiao, and Yingxia Shao. 2023. Making Large Language Models A Better Foundation For Dense Retrieval.CoRRabs/2312.15503 (2023). https://doi.org/10.48550/arXiv.2312.15503

    2023 doi

  11. [19]

    Dongyu Ru, Lin Qiu, Xiangkun Hu, Tianhang Zhang, Peng Shi, Shuaichen Chang, Cheng Jiayang, Cunxiang Wang, Shichao Sun, Huanyu Li, Zizhao Zhang, Binjie Wang, Jiarong Jiang, Tong He, Zhiguo Wang, Pengfei Liu, Yue Zhang, and Zheng Zhang. 2024. RAGChecker: A Fine-grained Framework...

    2024

  12. [20]

    Sparck Jones

    K. Sparck Jones. 1972. A Statistical Interpretation of Term Specificity and Its Application in Retrieval.Journal of Documentation28, 1 (1972), 11–21. https: //doi.org/10.1108/eb026526

    1972 doi

  13. [21]

    Yixuan Tang and Yi Yang. 2024. MultiHop-RAG: Benchmarking Retrieval- Augmented Generation for Multi-Hop Queries.CoRRabs/2401.15391 (2024). https://doi.org/10.48550/ARXIV.2401.15391 arXiv:2401.15391

    2024 doi arXiv

  14. [22]

    V. A. Traag, L. Waltman, and N. J. van Eck. 2019. From Louvain to Leiden: guaranteeing well-connected communities.Scientific Reports9, 1 (March 2019). https://doi.org/10.1038/s41598-019-41695-z

    2019 doi

  15. [23]

    Cohen, Ruslan Salakhutdinov, and Christopher D

    Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William W. Cohen, Ruslan Salakhutdinov, and Christopher D. Manning. 2018. HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering. InConference on Empirical Methods in Natural Language Processing (EMNLP). A...

    2018

  16. [2023]

    arXiv:2301.11792 [cs.CL] https://arxiv.org/abs/2301.11792

    Graph Attention with Hierarchies for Multi-hop Question Answering. arXiv:2301.11792 [cs.CL] https://arxiv.org/abs/2301.11792

    arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.