pith. sign in

arxiv: 2607.01852 · v1 · pith:AJ5RA3L2new · submitted 2026-07-02 · 💻 cs.IR · cs.AI· cs.CL

Evaluating Chunking Strategies for Retrieval-Augmented Generation on Academic Texts

Valentin J. J. Kreileder , Johannes Reisinger , Andreas Fischer This is my paper

Pith reviewed 2026-07-03 06:55 UTC · model grok-4.3

classification 💻 cs.IR cs.AIcs.CL
keywords chunking strategiesretrieval-augmented generationRAGsemantic chunkingRAGAsacademic textsevaluation framework
0
0 comments X

The pith

Cluster-based chunking did not outperform simpler fixed-size or recursive chunking for RAG on academic theses.

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

The paper tests whether cluster-based semantic chunking improves retrieval and answer quality in retrieval-augmented generation systems compared with fixed-size and recursive chunking. Experiments run on long structured academic theses and score outputs with the RAGAs framework. Results indicate no advantage for the cluster approach under the tested conditions. RAGAs faithfulness scores showed limited reliability, and performance varied sharply between fixed and document-specific questions, likely tied to document formatting and preprocessing.

Core claim

Under the tested configuration, cluster-based chunking did not outperform simpler strategies. RAGAs-based faithfulness shows limited reliability in this setup. Performance on fixed versus document-specific questions varied substantially, likely related to the formatting of documents and preprocessing.

What carries the argument

Direct comparison of three chunking methods (cluster-based semantic, fixed-size, recursive) inside a RAG pipeline, scored by RAGAs metrics on academic theses.

If this is right

  • Simpler chunking methods can be used without loss of quality for RAG on structured academic documents.
  • RAGAs faithfulness scores may not track human-perceived answer quality reliably on thesis-length texts.
  • Question design and document preprocessing choices affect measured RAG performance more than chunking method.
  • Fixed-size chunking remains a practical baseline for academic RAG applications.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Teams building RAG systems for academic content may save compute by defaulting to basic chunking unless domain-specific tests prove otherwise.
  • RAG evaluation frameworks require additional validation steps when applied to long-form technical documents.
  • The observed gap between fixed and document-specific questions points to a need for standardized test sets that control for document structure.

Load-bearing premise

The RAGAs framework supplies reliable measurements of faithfulness and answer quality for academic theses and the chosen question types.

What would settle it

Human raters scoring the same RAG answers for faithfulness and relevance, finding cluster-based chunking produces clearly superior results on the tested theses.

Figures

Figures reproduced from arXiv: 2607.01852 by Andreas Fischer, Johannes Reisinger, Valentin J. J. Kreileder.

Figure 1
Figure 1. Figure 1: Used chunking methods. Cluster-Based Chunking: Cluster-based chunking is a method where we combine semantically similar sentences to more coherent chunks ( [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 4
Figure 4. Figure 4: Example: Fixed question retrieval snippet Context F1 Results For fixed questions, context f1 medians are 0 across all chunkers. These first five questions target general information in the preliminaries. Even after cleaning, preliminary artifacts and dot leaders survive, polluting both indexing and retrieval. For free questions, context f1 medians reach approximately 0.5 for recursive chunking and 0.3 for … view at source ↗
Figure 2
Figure 2. Figure 2: TF-IDF cosine similarity with bigrams as terms. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Pipeline concerning the evaluation of chunking methods using RAGAS [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Boxplots displaying context F1 scores for the evaluated chunking strategies. [PITH_FULL_IMAGE:figures/full_fig_p004_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Boxplots displaying AQS scores for the evaluated chunking strategies. [PITH_FULL_IMAGE:figures/full_fig_p004_6.png] view at source ↗
read the original abstract

Retrieval-Augmented Generation (RAG) systems use the question-answering capabilities of Large Language Models (LLMs) to access information outside their parameters. We evaluate if cluster-based semantic chunking improves retrieval and answer quality compared to fixed-size and recursive chunking evaluating on long, structured academic theses using the Retrieval Augmented Generation Assessment (RAGAs) framework. RAGAs based faithfulness shows limited reliability in this setup. Performance on fixed versus document specific questions varied substantially, likely related to the formatting of documents and preprocessing. Under the tested configuration, cluster-based chunking did not outperform simpler strategies.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript evaluates cluster-based semantic chunking against fixed-size and recursive chunking for RAG on long academic theses, using the RAGAs framework to measure retrieval and answer quality. It reports that cluster-based chunking did not outperform the simpler strategies under the tested configuration, while noting that RAGAs faithfulness exhibits limited reliability and that performance varies substantially with document formatting and preprocessing.

Significance. If the comparative result can be placed on firmer evidential ground, the finding would indicate that semantic clustering adds little value over simpler chunking for structured academic documents, which could simplify RAG pipelines in scholarly settings. The work supplies a domain-specific empirical comparison on real theses, a useful data point given the length and hierarchical structure of such texts.

major comments (1)
  1. [Abstract] Abstract: The central claim that cluster-based chunking did not outperform simpler strategies is grounded in RAGAs faithfulness and answer-quality scores, yet the abstract itself states that 'RAGAs based faithfulness shows limited reliability in this setup'. Without independent validation (human evaluation, alternative metrics, or controls for the noted formatting effects), this self-identified limitation renders the 'did not outperform' conclusion insecure.
minor comments (2)
  1. [Abstract] Abstract: The statement that 'performance on fixed versus document specific questions varied substantially' is left without quantitative detail or statistical test; adding effect sizes or significance levels would strengthen the observation.
  2. The manuscript does not describe the exact question-generation procedure, the number of theses, or the LLM used for generation and evaluation; these omissions hinder reproducibility even if the RAGAs limitation is addressed.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on the abstract. We address the concern about the security of our central claim below and propose revisions to better qualify our findings.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that cluster-based chunking did not outperform simpler strategies is grounded in RAGAs faithfulness and answer-quality scores, yet the abstract itself states that 'RAGAs based faithfulness shows limited reliability in this setup'. Without independent validation (human evaluation, alternative metrics, or controls for the noted formatting effects), this self-identified limitation renders the 'did not outperform' conclusion insecure.

    Authors: We agree that the abstract's phrasing could more clearly signal the tentative nature of the finding given the metric limitation we ourselves identify. The manuscript already states both the empirical observation ('Under the tested configuration, cluster-based chunking did not outperform simpler strategies') and the caveat ('RAGAs based faithfulness shows limited reliability in this setup') in the same paragraph, and the results section further discusses substantial performance variation linked to document formatting and preprocessing. Nevertheless, to strengthen the presentation, we will revise the abstract to frame the result more explicitly as an observation within the RAGAs framework rather than a definitive comparative outcome, and we will add a brief clause noting that independent validation would be valuable for future work. This change preserves the paper's honest reporting of both the result and its limitations without overstating generalizability. revision: yes

Circularity Check

0 steps flagged

No circularity: pure empirical evaluation with no derivations or self-referential reductions

full rationale

The paper conducts an empirical comparison of chunking strategies (cluster-based, fixed-size, recursive) on academic theses, measuring outcomes via the external RAGAs framework. No equations, fitted parameters, uniqueness theorems, or derivations are present that could reduce to inputs by construction. The central claim rests on experimental results rather than any self-definitional or self-citation chain. Limitations in RAGAs reliability are explicitly noted but do not create circularity in the reported comparisons.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Empirical evaluation study; contains no free parameters, mathematical axioms, or invented entities.

pith-pipeline@v0.9.1-grok · 5629 in / 917 out tokens · 27037 ms · 2026-07-03T06:55:32.705219+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

73 extracted references · 50 canonical work pages · 13 internal anchors

  1. [1]

    and Raman, Marc M

    Mirsky, Grace M. and Raman, Marc M. , month = nov, year =. Data. 2025 3rd. doi:10.1109/FLLM67465.2025.11390957 , abstract =

    work page doi:10.1109/fllm67465.2025.11390957 2025

  2. [2]

    Campello, Ricardo J. G. B. and Moulavi, Davoud and Zimek, Arthur and Sander, Jörg , month = jul, year =. Hierarchical. doi:10.1145/2733381 , file =

    work page doi:10.1145/2733381

  3. [3]

    Campello, Ricardo J. G. B. and Moulavi, Davoud and Sander, Joerg , editor =. Density-. Advances in. 2013 , keywords =. doi:10.1007/978-3-642-37456-2_14 , abstract =

    work page doi:10.1007/978-3-642-37456-2_14 2013

  4. [4]

    Database Syst

    ACM Trans. Database Syst. , author =. 2017 , pages =. doi:10.1145/3068335 , abstract =

    work page doi:10.1145/3068335 2017

  5. [5]

    Wang, Yining and Wang, Liwei and Li, Yuanzhi and He, Di and Liu, Tie-Yan and Chen, Wei , month = apr, year =. A. doi:10.48550/arXiv.1304.6480 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1304.6480

  6. [6]

    Edge, Darren and Trinh, Ha and Cheng, Newman and Bradley, Joshua and Chao, Alex and Mody, Apurva and Truitt, Steven and Metropolitansky, Dasha and Ness, Robert Osazuwa and Larson, Jonathan , month = feb, year =. From. doi:10.48550/arXiv.2404.16130 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2404.16130

  7. [7]

    IEEE Transactions on Big Data , author =

    Billion-. IEEE Transactions on Big Data , author =. 2021 , keywords =. doi:10.1109/TBDATA.2019.2921572 , abstract =

    work page doi:10.1109/tbdata.2019.2921572 2021

  8. [8]

    Improved

    Ioffe, Sergey , month = dec, year =. Improved. 2010. doi:10.1109/ICDM.2010.80 , abstract =

    work page doi:10.1109/icdm.2010.80 2010

  9. [9]

    WIREs Data Mining and Knowledge Discovery , author =

    Algorithms for hierarchical clustering: an overview , volume =. WIREs Data Mining and Knowledge Discovery , author =. 2012 , pages =. doi:10.1002/widm.53 , abstract =

    work page doi:10.1002/widm.53 2012

  10. [10]

    and Matsuura, Kenji , year =

    Willmott, Cort J. and Matsuura, Kenji , year =. Advantages of the mean absolute error (. Climate Research , publisher =

  11. [11]

    The American Journal of Psychology 15, 72–101

    The. The American Journal of Psychology , author =. 1904 , pages =. doi:10.2307/1412159 , language =

    work page doi:10.2307/1412159 1904

  12. [12]

    scikit-learn , file =

  13. [13]

    Proceedings of

    Riedl, Martin and Biemann, Chris , editor =. Proceedings of. 2012 , pages =

    2012

  14. [14]

    arXiv:2402.05131 [cs.CL]

    Yepes, Antonio Jimeno and You, Yao and Milczek, Jan and Laverde, Sebastian and Li, Renyu , month = mar, year =. Financial. doi:10.48550/arXiv.2402.05131 , abstract =

    work page doi:10.48550/arxiv.2402.05131

  15. [15]

    Kreileder, Valentin and Fischer, Andreas and Reisinger, Johannes , month = jan, year =. Thesis-

  16. [16]

    2026 , note =

    vibrantlabsai/ragas , copyright =. 2026 , note =

    2026

  17. [17]

    and Raghavan, Prabhakar and Schütze, Hinrich , month = jul, year =

    Manning, Christopher D. and Raghavan, Prabhakar and Schütze, Hinrich , month = jul, year =. Introduction to. Cambridge Aspire website , publisher =. doi:10.1017/CBO9780511809071 , note =

    work page doi:10.1017/cbo9780511809071

  18. [18]

    Document

    Wang, Zhitong and Gao, Cheng and Xiao, Chaojun and Huang, Yufei and Si, Shuzheng and Luo, Kangyang and Bai, Yuzhuo and Li, Wenhao and Duan, Tangjian and Lv, Chuancheng and Lu, Guoshan and Chen, Gang and Qi, Fanchao and Sun, Maosong , editor =. Document. Findings of the. 2025 , pages =. doi:10.18653/v1/2025.findings-acl.422 , abstract =

    work page doi:10.18653/v1/2025.findings-acl.422 2025

  19. [19]

    ACM Comput

    A. ACM Comput. Surv. , author =. 2023 , pages =. doi:10.1145/3606367 , abstract =

    work page doi:10.1145/3606367 2023

  20. [20]

    1973 , pages =

    The Computer Journal , author =. 1973 , pages =. doi:10.1093/comjnl/16.1.30 , abstract =

    work page doi:10.1093/comjnl/16.1.30 1973

  21. [21]

    Qu, Renyi and Tu, Ruixuan and Bao, Forrest Sheng , editor =. Is. Findings of the. 2025 , pages =. doi:10.18653/v1/2025.findings-naacl.114 , abstract =

    work page doi:10.18653/v1/2025.findings-naacl.114 2025

  22. [22]

    Evaluation of

    Yu, Hao and Gan, Aoran and Zhang, Kai and Tong, Shiwei and Liu, Qi and Liu, Zhaofeng , year =. Evaluation of. doi:10.1007/978-981-96-1024-2_8 , abstract =

    work page doi:10.1007/978-981-96-1024-2_8

  23. [23]

    Proceedings of the 18th

    Es, Shahul and James, Jithin and Espinosa Anke, Luis and Schockaert, Steven , editor =. Proceedings of the 18th. 2024 , pages =. doi:10.18653/v1/2024.eacl-demo.16 , abstract =

    work page doi:10.18653/v1/2024.eacl-demo.16 2024

  24. [24]

    Information Storage and Retrieval , author =

    A definition of relevance for information retrieval , volume =. Information Storage and Retrieval , author =. 1971 , pages =. doi:10.1016/0020-0271(71)90024-6 , abstract =

    work page doi:10.1016/0020-0271(71)90024-6 1971

  25. [25]

    How to split text based on semantic similarity

  26. [26]

    Chase, Harrison , month = oct, year =

  27. [27]

    Günther, Michael and Mohr, Isabelle and Williams, Daniel James and Wang, Bo and Xiao, Han , month = jul, year =. Late. doi:10.48550/arXiv.2409.04701 , abstract =

    work page doi:10.48550/arxiv.2409.04701

  28. [28]

    Rethinking

    Bhat, Sinchana Ramakanth and Rudat, Max and Spiekermann, Jannis and Flores-Herr, Nicolas , month = may, year =. Rethinking. doi:10.48550/arXiv.2505.21700 , abstract =

    work page doi:10.48550/arxiv.2505.21700

  29. [29]

    IEEE Transactions on Pattern Analysis and Machine Intelligence , author =

    n-. IEEE Transactions on Pattern Analysis and Machine Intelligence , author =. 1979 , keywords =. doi:10.1109/TPAMI.1979.4766902 , abstract =

    work page doi:10.1109/tpami.1979.4766902 1979

  30. [30]

    Étude comparative de la distribution florale dans une portion des

    Jaccard, Paul , year =. Étude comparative de la distribution florale dans une portion des. Bulletin de la Société Vaudoise des Sciences Naturelles , publisher =. doi:10.5169/seals-266450 , number =

    work page doi:10.5169/seals-266450

  31. [31]

    Procedia Computer Science , author =

    Matching. Procedia Computer Science , author =. 2024 , keywords =. doi:10.1016/j.procs.2024.03.039 , abstract =

    work page doi:10.1016/j.procs.2024.03.039 2024

  32. [32]

    InProceedings of the 5th Workshop on Research in Computational Linguistic Typology and Multilingual NLP, pages 89–95, Dubrovnik, Croatia

    Nussbaum, Zach and Duderstadt, Brandon , month = mar, year =. Training. doi:10.48550/arXiv.2502.07972 , abstract =

    work page doi:10.48550/arxiv.2502.07972

  33. [33]

    Thakur, Nandan and Reimers, Nils and Rücklé, Andreas and Srivastava, Abhishek and Gurevych, Iryna , month = aug, year =

  34. [34]

    Caspari, Laura and Dastidar, Kanishka Ghosh and Zerhoudi, Saber and Mitrovic, Jelena and Granitzer, Michael , month = jul, year =. Beyond. doi:10.48550/arXiv.2407.08275 , abstract =

    work page doi:10.48550/arxiv.2407.08275

  35. [35]

    Measurement of

    Wang, Jiapeng and Dong, Yihong , month = sep, year =. Measurement of. Information , publisher =. doi:10.3390/info11090421 , abstract =

    work page doi:10.3390/info11090421

  36. [36]

    Chroma , file =

    Chroma , url =. Chroma , file =

  37. [37]

    Engineering at Meta , month = mar, year =

    Faiss:. Engineering at Meta , month = mar, year =

  38. [38]

    The vector database to build knowledgeable

  39. [39]

    Cohen, Ruslan Salakhut- dinov, and Christopher D

    Yang, Zhilin and Qi, Peng and Zhang, Saizheng and Bengio, Yoshua and Cohen, William and Salakhutdinov, Ruslan and Manning, Christopher D. , editor =. Proceedings of the 2018. 2018 , pages =. doi:10.18653/v1/D18-1259 , abstract =

    work page doi:10.18653/v1/d18-1259 2018

  40. [40]

    2025 , note =

    allenai/scidocs , url =. 2025 , note =

    2025

  41. [41]

    Zhang, Yue and Li, Yafu and Cui, Leyang and Cai, Deng and Liu, Lemao and Fu, Tingchen and Huang, Xinting and Zhao, Enbo and Zhang, Yu and Xu, Chen and Chen, Yulong and Wang, Longyue and Luu, Anh Tuan and Bi, Wei and Shi, Freda and Shi, Shuming , month = sep, year =. Siren's. doi:10.48550/arXiv.2309.01219 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2309.01219

  42. [42]

    Zhang, Yanzhao and Li, Mingxin and Long, Dingkun and Zhang, Xin and Lin, Huan and Yang, Baosong and Xie, Pengjun and Yang, An and Liu, Dayiheng and Lin, Junyang and Huang, Fei and Zhou, Jingren , month = jun, year =. Qwen3. doi:10.48550/arXiv.2506.05176 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2506.05176

  43. [43]

    2024 , file =

    nomic-ai/nomic-embed-text-v1 ·. 2024 , file =

    2024

  44. [44]

    Arize AI , file =

    Understanding and. Arize AI , file =

  45. [45]

    Analytics Vidhya , author =

    8. Analytics Vidhya , author =. 2025 , file =

    2025

  46. [46]

    Zhong, Zijie and Liu, Hanwen and Cui, Xiaoya and Zhang, Xiaofan and Qin, Zengchang , month = jun, year =. Mix-of-. doi:10.48550/arXiv.2406.00456 , abstract =

    work page doi:10.48550/arxiv.2406.00456

  47. [47]

    IEEE Transactions on Knowledge and Data Engineering , author =

    Neural. IEEE Transactions on Knowledge and Data Engineering , author =. 2022 , keywords =. doi:10.1109/TKDE.2020.2983360 , abstract =

    work page doi:10.1109/tkde.2020.2983360 2022

  48. [48]

    and McKeown, Kathleen

    Liu, Nelson F. and Lin, Kevin and Hewitt, John and Paranjape, Ashwin and Bevilacqua, Michele and Petroni, Fabio and Liang, Percy , month = nov, year =. Lost in the. doi:10.48550/arXiv.2307.03172 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2307.03172

  49. [49]

    https://openaccess.nhh.no/nhh-xmlui/bitstream/handle/11250/3178510/no.nhh\

  50. [50]

    BERTopic: Neural topic modeling with a class-based TF-IDF procedure

    Grootendorst, Maarten , month = mar, year =. doi:10.48550/arXiv.2203.05794 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2203.05794

  51. [51]

    https://simg.baai.ac.cn/paperfile/25a43194-c74c-4cd3-b60f-0a1f27f8b8af.pdf , url =

  52. [52]

    Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks

    Lewis, Patrick and Perez, Ethan and Piktus, Aleksandra and Petroni, Fabio and Karpukhin, Vladimir and Goyal, Naman and Küttler, Heinrich and Lewis, Mike and Yih, Wen-tau and Rocktäschel, Tim and Riedel, Sebastian and Kiela, Douwe , month = apr, year =. Retrieval-. doi:10.48550/arXiv.2005.11401 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2005.11401 2005

  53. [53]

    https://arxiv.org/pdf/2005.11401 , url =

    work page internal anchor Pith review Pith/arXiv arXiv 2005

  54. [54]

    Retrieval-Augmented Generation for Large Language Models: A Survey

    Gao, Yunfan and Xiong, Yun and Gao, Xinyu and Jia, Kangxiang and Pan, Jinliu and Bi, Yuxi and Dai, Yi and Sun, Jiawei and Wang, Meng and Wang, Haofen , month = mar, year =. Retrieval-. doi:10.48550/arXiv.2312.10997 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2312.10997

  55. [55]

    Bavarian Journal of Applied Sciences , author =

    Topical. Bavarian Journal of Applied Sciences , author =. 2023 , note =. doi:10.25929/1rjp-d197 , abstract =

    work page doi:10.25929/1rjp-d197 2023

  56. [56]

    MTEB: Massive Text Embedding Benchmark

    Muennighoff, Niklas and Tazi, Nouamane and Magne, Loïc and Reimers, Nils , month = mar, year =. doi:10.48550/arXiv.2210.07316 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2210.07316

  57. [57]

    Retrieval-based

    Asai, Akari and Min, Sewon and Zhong, Zexuan and Chen, Danqi , editor =. Retrieval-based. Proceedings of the 61st. 2023 , pages =. doi:10.18653/v1/2023.acl-tutorials.6 , abstract =

    work page doi:10.18653/v1/2023.acl-tutorials.6 2023

  58. [58]

    BERT : Pre-training of Deep Bidirectional Transformers for Language Understanding

    Devlin, Jacob and Chang, Ming-Wei and Lee, Kenton and Toutanova, Kristina , editor =. Proceedings of the 2019. 2019 , pages =. doi:10.18653/v1/N19-1423 , abstract =

    work page doi:10.18653/v1/n19-1423 2019

  59. [59]

    Contextual

    Bari, Harsh , month = jan, year =. Contextual. 2025. doi:10.1109/SCEECS64059.2025.10940513 , abstract =

    work page doi:10.1109/sceecs64059.2025.10940513 2025

  60. [60]

    Putra, Syopiansyah Jaya and Gunawan, Muhamad Nur and Hidayat, Arief Akbar , month = sep, year =. Feature. 2022 10th. doi:10.1109/CITSM56380.2022.9935873 , abstract =

    work page doi:10.1109/citsm56380.2022.9935873 2022

  61. [61]

    Robust and

    Rahman, Rifat , month = dec, year =. Robust and. 2020 23rd. doi:10.1109/ICCIT51783.2020.9392738 , abstract =

    work page doi:10.1109/iccit51783.2020.9392738 2020

  62. [62]

    Wang, Liang and Yang, Nan and Huang, Xiaolong and Jiao, Binxing and Yang, Linjun and Jiang, Daxin and Majumder, Rangan and Wei, Furu , month = feb, year =. Text. doi:10.48550/arXiv.2212.03533 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2212.03533

  63. [63]

    Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

    Reimers, Nils and Gurevych, Iryna , month = aug, year =. Sentence-. doi:10.48550/arXiv.1908.10084 , abstract =

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1908.10084 1908

  64. [64]

    Semantic

    Reisinger, Johannes and Fischer, Andreas and Igl, Andreas , month = aug, year =. Semantic. 2025 2nd. doi:10.1109/GACLM67198.2025.11231968 , abstract =

    work page doi:10.1109/gaclm67198.2025.11231968 2025

  65. [65]

    Advances in Neural Information Processing Systems , author =

    Judging. Advances in Neural Information Processing Systems , author =. 2023 , pages =

    2023

  66. [66]

    2024 , keywords =

    sentence-transformers/all-. 2024 , keywords =

    2024

  67. [67]

    Y. Gao, Y. Xiong, X. Gao, K. Jia, J. Pan, Y. Bi, Y. Dai, J. Sun, M. Wang, and H. Wang, `` en Retrieval- Augmented Generation for Large Language Models : A Survey ,'' Mar. 2024, arXiv:2312.10997 [cs]. [Online]. Available: http://arxiv.org/abs/2312.10997

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  68. [68]

    R. Qu, R. Tu, and F. S. Bao, ``Is Semantic Chunking Worth the Computational Cost ?'' in Findings of the Association for Computational Linguistics : NAACL 2025 , L. Chiruzzo, A. Ritter, and L. Wang, Eds. 1em plus 0.5em minus 0.4em Albuquerque, New Mexico: Association for Computational Linguistics, Apr. 2025, pp. 2155--2177. [Online]. Available: https://acl...

    2025

  69. [69]

    Günther, I

    M. Günther, I. Mohr, D. J. Williams, B. Wang, and H. Xiao, ``Late Chunking : Contextual Chunk Embeddings Using Long - Context Embedding Models ,'' Jul. 2025, arXiv:2409.04701 [cs]. [Online]. Available: http://arxiv.org/abs/2409.04701

    work page arXiv 2025

  70. [70]

    Reisinger, A

    J. Reisinger, A. Fischer, and A. Igl, ``Semantic Document Graphs for Knowledge Retrieval ,'' in 2025 2nd International Generative AI and Computational Language Modelling Conference ( GACLM ) , Aug. 2025, pp. 294--298. [Online]. Available: https://ieeexplore.ieee.org/document/11231968

    work page arXiv 2025

  71. [71]

    ``sentence-transformers/all- MiniLM - L6 -v2 · Hugging Face ,'' Jan. 2024. [Online]. Available: https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2

    2024

  72. [72]

    S. Es, J. James, L. Espinosa Anke, and S. Schockaert, `` RAGAs : Automated Evaluation of Retrieval Augmented Generation ,'' in Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics : System Demonstrations , N. Aletras and O. De Clercq, Eds. 1em plus 0.5em minus 0.4em St. Julians, Malta: Association for...

    2024

  73. [73]

    C. D. Manning, P. Raghavan, and H. Schütze, `` en Introduction to Information Retrieval ,'' Jul. 2008, iSBN: 9780511809071. [Online]. Available: https://www.cambridge.org/highereducation/books/introduction-to-information-retrieval/669D108D20F556C5C30957D63B5AB65C

    2008