Pith. sign in

REVIEW 4 major objections 6 minor 42 references

Multi-Facet Blending for Faceted Query-by-Example Retrieval

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read Faceted query-by-example retrieval can be trained without facet-level labels by having a language model decompose documents, generate facet-specific similar and dissimilar fragments, and recompose them into synthetic training triples.

desk verdict Solid augmentation recipe and a useful new benchmark, but the facet-conditional claim rests on an untested sign of LLM-generated triplets. read the letter →

arxiv 2412.01443 v1 pith:GWXATTUC submitted 2024-12-02 cs.IR cs.AIcs.CL

classification cs.IRcs.AIcs.CL
keywords facetedquery-by-exampledataaugmentationLLM-generatedtrainingcontrastivelearningscientificpaperretrievaleducationalitemfacetdecompositionbenchmark
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

The paper claims that fine-grained 'query by example' retrieval—finding documents similar to a query document under a specified facet such as background, method, or result—can be trained without any facet-level relevance labels. The proposed FaBle method lets an open language model decompose each document into facet summaries, generate facet-matched and facet-mismatched fragments, and recompose those fragments into synthetic positive and negative training documents. Fine-tuning a document encoder on these synthetic pairs improves retrieval on the CSFCube scientific-paper benchmark, with the largest gains on the method facet, and also improves retrieval on a new educational exam-item benchmark, FEIR, introduced by the paper. The paper's bottom line is that modular synthesis, rather than citation labels or human annotations, can supply the facet signal that current methods lack.

What carries the argument

The load-bearing mechanism is multi-facet blending: a three-stage pipeline that decomposes (prompts LLaMA2-13B to summarize each facet), generates (self-feeds the facet summary into prompts asking for a similar and a dissimilar facet fragment), and recomposes (concatenates the generated fragments to form facet-conditioned positives and negatives, and optionally rescores negatives with a MiniLM cross-encoder to create hard negatives). This machinery turns one unlabeled document into forty triplet pairs per facet, so roughly 1K abstracts yield 40.68K training triples.

What would settle it

Take the 1K abstracts, have human annotators mark which sentences realize each facet, and measure overlap with LLaMA2's generated summaries; low agreement would mean the synthetic pairs encode the model's bias rather than facet-level relevance, so the reported gains would not transfer to new domains.

Watch

Extended reading notes

Core claim

The central discovery is that explicit facet-conditioned training pairs can be synthesized without labels: a small open LLM (LLaMA2-13B) summarizes each facet of an abstract, generates a similar version of that facet and a dissimilar version on a different topic, and the recomposed combinations serve as positive and negative documents for contrastive training. On CSFCube, applying this augmentation to SPECTER yields average gains of 3.4 NDCG@20 and 1.4 MAP, with the method facet gaining 7.6 NDCG@20 and 3.5 MAP; with the co-citation-trained SPECTER-COCITE base, FaBle outperforms ASPIRE models trained on roughly 32 times more data. The paper also claims domain transfer: on FEIR, built from TOEFL listening items, every facet improves after FaBle augmentation.

Load-bearing premise

The whole method rests on the assumption that the language model's automatic summaries and generated snippets isolate the requested facet the way a human would, because those snippets are the only training signal.

Editorial extensions

If this is right

  • Fine-tuning on FaBle pairs improves every facet of CSFCube over SPECTER, with the largest effect on the method facet, which prior models handle worst.
  • The same augmentation transfers to a new domain: on the released FEIR benchmark built from TOEFL items, all facets improve, with the largest gains on the question facet.
  • Because the pipeline needs no citations or facet labels, it opens faceted QBE to domains like education and law where such labels do not exist.
  • Increasing the FaBle-augmented corpus from 0.5K to 1K documents consistently raises NDCG@20 and MAP, especially for the method facet.
  • Hard negatives generated by score-controlled regeneration help the method facet but not background or result, suggesting the augmentation can be tailored per facet.

Reading between the lines

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

  • If the facet summaries are faithful, the same decomposition could serve as a reusable pre-training signal: any corpus with a named facet taxonomy could be augmented without human labels, not only abstracts and exam items.
  • The hard-negative rescoring step suggests a testable extension: replacing the MiniLM scorer with a prompt asking the LLM to self-score generated fragments might make the pipeline fully self-contained.
  • A direct comparison against training on human-annotated facet pairs of the same size would isolate how much of the gain comes from the synthesis itself rather than from the contrastive objective.
  • The FEIR benchmark, with eight queries per facet, is small, yet it gives the education domain a first public testbed for facet-conditioned retrieval.
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

4 major / 6 minor

Summary. The paper introduces FaBle, a fully synthetic augmentation method for faceted query-by-example retrieval. The method decomposes each document into facet summaries using zero-shot prompting of LLaMA2-13B, generates facet-specific 'similar' and 'dissimilar' fragments by recursive prompting, and recomposes these fragments into positive and negative pseudo-documents. The authors fine-tune SPECTER on the resulting triplet pairs and evaluate on CSFCube, where they report aggregate NDCG%20 and MAP gains over the base SPECTER, and on a newly released FEIR benchmark for TOEFL-style exam items, where gains are reported for all facets. The paper also includes an ablation replacing generated negatives with random facets and a hard-negative regeneration step based on MiniLM scores.

Significance. If the reported gains hold up, the paper's contribution is a low-resource route to facet-conditional embeddings: it uses only 1K documents and no citation labels or manual facet annotations, and it provides the first faceted QBE test set in the education domain. The release of code, datasets, and the FEIR benchmark, together with the query-level win-rate analysis and random-negative ablation, are concrete strengths. The main caveat is that the central causal claim—that LLM similarity judgments teach true facet semantics—rests on an assumption that is asserted but not validated; the label-sign is never tested against an independent signal. The absence of deduplication between the S2ORC training sample and the CSFCube test set and the score-variance-based selection of FEIR queries add further uncertainty.

major comments (4)
  1. [§3.2, Eqs. (3)–(4); §7, Table 4] The label-sign of the LLM-generated positives and negatives is never independently validated. The only distinction between D_f+ and D_f- is the prompt phrasing 'write a similar {facet}' versus 'write a dissimilar {facet}', both conditioned on the same LLaMA2-13B summary S_f. The FaBle-RN ablation in Table 4 replaces only the negative fragments with random original facets; it does not swap the labels or replace C_f^sim with a random facet. Consequently, the observed gains are also consistent with generic in-domain triplet fine-tuning on 40K syntactically coherent paper-text pairs, rather than with the facet-conditional semantics claimed in the abstract. I ask for two control conditions: (i) train with the labels flipped (C_f^dis as positives and C_f^sim as negatives), and (ii) train with random original facet summaries as positives while keeping the generated negatives; if either control also improves over the SPECTER baseline, the causal interpretation in the abstract is not supported. In addition, the assertion in §3.1 that S_f 'highly represents the facet' is supported only by qualitative examples in Figure 3 and by full-document similarity in Table 3, neither of which measures facet fidelity against a reference.
  2. [§5, Data and Settings] The 1,017 S2ORC abstracts are selected from the computer science domain, the same domain as the CSFCube test collection, but the paper does not report whether these abstracts are deduplicated against the CSFCube query or candidate papers. Because FaBle constructs pseudo-documents from the full text of these abstracts, any overlap between the augmentation source and the test collection could inflate the reported gains. Please report the exact number of overlapping paper identifiers between the S2ORC sample and the CSFCube papers, and if the count is nonzero, re-run the experiments after removing the overlapping abstracts.
  3. [§4, Dataset Construction] The FEIR test queries are selected by taking the eight items with the largest standard deviations of MiniLM similarity scores, and the same MiniLM cross-encoder is used in the hard-negative generation pipeline of §3.5. This means the new benchmark is intentionally biased toward items that MiniLM-style representations find heterogeneous, which may make the test set unrepresentative of normal exam-item retrieval and may align with the inductive biases of the augmentation procedure. Please justify this selection criterion with respect to the intended use of FEIR, and report results under a random selection of queries or across multiple selection seeds to show the benchmark is not tuned to the method.
  4. [Limitations; Tables 2 and 5] The paper does not provide significance tests, and the Limitations section states that no statistical tests are provided due to the small test-set size. Several facet-level comparisons move in the opposite direction (e.g., Background MAP decreases from 43.95 to 42.66 in Table 2), and the standard deviations overlap for some gains. Since the abstract uses 'remarkably' and the text uses 'significantly' for the method facet, the central claim should be supported by bootstrap confidence intervals or paired significance tests across the 50 CSFCube and 24 FEIR queries, or by clearly presenting the query-level win rates already reported in Limitations as the main robustness evidence.
minor comments (6)
  1. [§1, p. 2] 'Fable' is a typo for 'FaBle', and the sentence 'Fable explicitly create' has a subject-verb agreement error.
  2. [Table 2] The header 'CFSCUBE' should be 'CSFCube'.
  3. [§3.3] The pair-count derivation is unclear: after stating that four D_f+ and four D_f- are generated per document, the text says 'five documents, including the original one, lead to ten (D_f;Q, D_f+) pairs (i.e., five choose two)', which does not specify which documents are anchors and which are positives. Please rewrite the counting argument with explicit per-facet totals.
  4. [§5, Evaluation Metrics; Appendix D] The text says NDCG%20 is computed at 20% of the query pool size, but the formula K = p*|C| uses the candidate pool size; please correct the wording.
  5. [Ethical Statement] The Ethical Statement says 'two human experts', while §4 says three experts were hired (a professor and two English specialists); please reconcile the numbers.
  6. [Figure 2] The prompt text in Figure 2 is too small to read in the PDF; please provide the full prompts in a table or appendix.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FaBle's synthetic training pairs are evaluated on external human-annotated benchmarks, so the central claim is not reduced to its own inputs.

full rationale

The paper's derivation chain is self-contained rather than circular. FaBle constructs facet-conditional triplet pairs via LLM-based decomposition and generation (Eqs. 1-4), then fine-tunes SPECTER with contrastive loss. The reported gains are measured on CSFCube, an existing external benchmark, and on FEIR, a newly released test set whose relevance labels were provided by independent human experts (Section 4, 'Relevance Annotation'). No parameter is fitted to the test labels and then reported as a prediction: the MiniLM rescoring in Section 3.5 is used only to regenerate hard negatives, not to define the headline metrics. The only self-citations (e.g., Do and Lee 2024 in Section 2) are background references and are not load-bearing for the main result. The skeptical concern that the LLM-generated positive/negative signs are unvalidated is a legitimate correctness or assumption-risk issue, but it is not a circularity issue: even if the LLM's judgments were noisy, the training signal would still be externally evaluated against human judgments, so the claim could in principle fail. No equation or fitted value reduces to the target result by construction.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The method is an empirical pipeline with several hand-chosen thresholds and strong assumptions about unlabeled documents. It introduces no new theoretical entities; its load-bearing premises are that documents have identifiable facets, that the LLM's decomposition and generation reflect those facets, and that MiniLM pseudo-relevance scores are a valid guide for hard negatives.

free parameters (3)
  • MiniLM hard-negative score thresholds = 0.25, 0.5
    Hand-chosen cutoffs in Section 3.5 determine which generated dissimilar facets are regenerated and accepted as hard negatives; these thresholds are not justified by data or ablations beyond the final result.
  • Triplet loss margin m = 1
    Set in Appendix B without sensitivity analysis; affects contrastive training but is a standard choice.
  • Facet set size and names = 3 facets: background, method, result; story, question, options for FEIR
    The number of facets and their names are fixed by the user or prompt, not learned; the 'no pre-defined facet knowledge' claim depends on this being provided.
assumptions (5)
  • domain assumption Documents are composed of distinct facets (background/method/result for abstracts; story/question/options for exam items) that can be summarized and recombined independently.
    Used throughout Section 3; the existence and independence of these facets is assumed, not learned or validated.
  • domain assumption Zero-shot LLaMA2-13B summarization extracts the intended facet faithfully.
    Section 3.1 states the summary 'highly represents the facet' with only qualitative support (Figure 3).
  • domain assumption LLM-generated similar and dissimilar facet fragments are valid proxies for facet-level relevance in training.
    Equations (1) and (2) define the training signal; no human comparison is used during generation.
  • ad hoc to paper MiniLM similarity scores are a valid pseudo-relevance signal for defining hard negatives.
    Section 3.5 applies hand-chosen thresholds (0.25, 0.5) to MiniLM scores without evidence they align with human relevance.
  • domain assumption The 1K sampled abstracts do not overlap with the CSFCube test set.
    No deduplication step is reported between the S2ORC sample and CSFCube queries or candidates.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Facet Blending for Faceted Query-by-Example Retrieval." pith.science (2026). https://pith.science/paper/GWXATTUC

@misc{pith2026241201443,
  author       = {Pith},
  title        = {Pith review of: Multi-Facet Blending for Faceted Query-by-Example Retrieval},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GWXATTUC}},
  note         = {Machine review of arXiv:2412.01443}
}
read the original abstract

With the growing demand to fit fine-grained user intents, faceted query-by-example (QBE), which retrieves similar documents conditioned on specific facets, has gained recent attention. However, prior approaches mainly depend on document-level comparisons using basic indicators like citations due to the lack of facet-level relevance datasets; yet, this limits their use to citation-based domains and fails to capture the intricacies of facet constraints. In this paper, we propose a multi-facet blending (FaBle) augmentation method, which exploits modularity by decomposing and recomposing to explicitly synthesize facet-specific training sets. We automatically decompose documents into facet units and generate (ir)relevant pairs by leveraging LLMs' intrinsic distinguishing capabilities; then, dynamically recomposing the units leads to facet-wise relevance-informed document pairs. Our modularization eliminates the need for pre-defined facet knowledge or labels. Further, to prove the FaBle's efficacy in a new domain beyond citation-based scientific paper retrieval, we release a benchmark dataset for educational exam item QBE. FaBle augmentation on 1K documents remarkably assists training in obtaining facet conditional embeddings.

Figures

Figures reproduced from arXiv: 2412.01443 by the authors.

Figure 1
Figure 1. Examples of documents with multiple facets. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of the FaBle method and examples of detailed prompts used for scientific paper retrieval. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Examples of the generated similar and dissimilar method facets with our self-fed decomposition (left) and without the (w/o) decomposition (right). Directly generating similar and dissimilar facets without decomposition can lead to the results containing facets other than the intended one, as highlighted. five choose two, 5 2  ). For each of them, one Df− is selected among four candidates, resulting in a total of fo… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Hard negative generation procedure (§ 3.5). Thus, to enhance the FaBle-generated facets from the perspective of negative sampling, we em￾ploy MiniLM2 (Wang et al., 2020), a lightweight cross-encoder model trained on MS MARCO (Ba￾jaj et al., 2016) using knowledge distil…
Figure 5
Figure 5. Figure 5: Score label distributions per query by facet. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 7
Figure 7. Figure 7: Comparison of NDCG%20 (left) and MAP (right) performances by the dataset size per facet. facets, explaining the less pronounced impact of our facet-specific approach on it. In the result facet, the impact of FaBle is evident, although not as large as the method. This o…
Figure 6
Figure 6. Figure 6: MiniLM score distributions of generated Sim and Dis facets. (a) Initial distributions; (b) shifts in negative samples after regeneration; dashed line: mean. Unlike them, FaBle leverages the knowledge em￾bedded within LLMs trained on massive corpora to make intrinsic ju…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 18 canonical work pages

  1. [1]

    online" 'onlinestring :=

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint eprinttype howpublished institution journal key month note number organization pages publisher school series title type volume year doi pubmed url lastchecked label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block STRING...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Chittaranjan Andrade. 2011. How to write a good abstract for a scientific paper or conference presentation. Indian journal of psychiatry, 53(2):172

  4. [4]

    Payal Bajaj, Daniel Campos, Nick Craswell, Li Deng, Jianfeng Gao, Xiaodong Liu, Rangan Majumder, Andrew McNamara, Bhaskar Mitra, Tri Nguyen, et al. 2016. Ms marco: A human generated machine reading comprehension dataset. arXiv preprint arXiv:1611.09268

  5. [5]

    Iz Beltagy, Kyle Lo, and Arman Cohan. 2019. https://doi.org/10.18653/v1/D19-1371 S ci BERT : A pretrained language model for scientific text . In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3615--3620, Hong Kong, China...

  6. [6]

    Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin...

  7. [7]

    Cheng-Han Chiang and Hung-yi Lee. 2023. Can large language models be an alternative to human evaluations? arXiv preprint arXiv:2305.01937

  8. [8]

    Yu-An Chung, Hung-Yi Lee, and James Glass. 2018. Supervised and unsupervised transfer learning for question answering. In NAACL HLT

Show all 42 references
  1. [9]

    Arman Cohan, Sergey Feldman, Iz Beltagy, Doug Downey, and Daniel Weld. 2020. https://doi.org/10.18653/v1/2020.acl-main.207 SPECTER : Document-level representation learning using citation-informed transformers . In Proceedings of the 58th Annual Meeting of the Association for C...

  2. [10]

    Heejin Do and Gary Geunbae Lee. 2024. Aspect-based semantic textual similarity for educational test items. In International Conference on Artificial Intelligence in Education, pages 344--352. Springer

  3. [11]

    Cody Dunne, Ben Shneiderman, Robert Gove, Judith Klavans, and Bonnie Dorr. 2012. Rapid understanding of scientific paper collections: Integrating statistics, text analytics, and visualization. Journal of the American Society for Information Science and Technology, 63(12):2351--2369

  4. [12]

    Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and St \'e phane Clinchant. 2022. From distillation to hard negative sampling: Making sparse neural ir models more effective. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Inf...

  5. [13]

    Luyu Gao, Xueguang Ma, Jimmy Lin, and Jamie Callan. 2023. https://doi.org/10.18653/v1/2023.acl-long.99 Precise zero-shot dense retrieval without relevance labels . In Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers...

  6. [14]

    Tom Hope, Jason Portenoy, Kishore Vasan, Jonathan Borchardt, Eric Horvitz, Daniel Weld, Marti Hearst, and Jevin West. 2020. https://doi.org/10.18653/v1/2020.emnlp-demos.18 S ci S ight: Combining faceted navigation and research group detection for COVID -19 exploratory scientif...

  7. [15]

    Joonseok Lee, Kisung Lee, and Jennifer G Kim. 2013. Personalized academic research paper recommendation system. arXiv preprint arXiv:1304.5457

  8. [16]

    Haitao Li, Weihang Su, Changyue Wang, Yueyue Wu, Qingyao Ai, and Yiqun Liu. 2023. Thuir@ coliee 2023: Incorporating structural knowledge into pre-trained language models for legal case retrieval. arXiv preprint arXiv:2305.06812

  9. [17]

    Matteo Lissandrini, Davide Mottin, Themis Palpanas, and Yannis Velegrakis. 2019. Example-based search: a new frontier for exploratory search. In Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 1411--1412

  10. [18]

    Kyle Lo, Lucy Lu Wang, Mark Neumann, Rodney Kinney, and Daniel Weld. 2020. https://doi.org/10.18653/v1/2020.acl-main.447 S 2 ORC : The semantic scholar open research corpus . In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 4969...

  11. [19]

    Kelvin Luu, Xinyi Wu, Rik Koncel-Kedziorski, Kyle Lo, Isabel Cachola, and Noah A. Smith. 2021. https://doi.org/10.18653/v1/2021.acl-long.166 Explaining relationships between scientific documents . In Proceedings of the 59th Annual Meeting of the Association for Computational L...

  12. [20]

    Sheshera Mysore, Arman Cohan, and Tom Hope. 2022. https://doi.org/10.18653/v1/2022.naacl-main.331 Multi-vector models with textual guidance for fine-grained scientific document similarity . In Proceedings of the 2022 Conference of the North American Chapter of the Association ...

  13. [21]

    Sheshera Mysore, Tim O'Gorman, Andrew McCallum, and Hamed Zamani. 2021. Csfcube-a test collection of computer science research articles for faceted query by example. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track (Round 2)

  14. [22]

    Mariana Neves, Daniel Butzke, and Barbara Grune. 2019. Evaluation of scientific elements for text similarity in biomedical publications. In Proceedings of the 6th Workshop on Argument Mining, pages 124--135

  15. [23]

    OpenAI. 2023. https://arxiv.org/abs/2303.08774 Gpt-4 technical report . Preprint, arXiv:2303.08774

  16. [24]

    Malte Ostendorff, Nils Rethmeier, Isabelle Augenstein, Bela Gipp, and Georg Rehm. 2022. https://doi.org/10.18653/v1/2022.emnlp-main.802 Neighborhood contrastive learning for scientific document representations with citation embeddings . In Proceedings of the 2022 Conference on...

  17. [25]

    Malte Ostendorff, Terry Ruas, Till Blume, Bela Gipp, and Georg Rehm. 2020 a . https://doi.org/10.18653/v1/2020.coling-main.545 Aspect-based document similarity for research papers . In Proceedings of the 28th International Conference on Computational Linguistics, pages 6194--6...

  18. [26]

    Malte Ostendorff, Terry Ruas, Moritz Schubotz, Georg Rehm, and Bela Gipp. 2020 b . https://doi.org/10.1145/3383583.3398525 Pairwise multi-class document classification for semantic relations between wikipedia articles . In Proceedings of the ACM/IEEE Joint Conference on Digita...

  19. [27]

    Yingqi Qu, Yuchen Ding, Jing Liu, Kai Liu, Ruiyang Ren, Wayne Xin Zhao, Daxiang Dong, Hua Wu, and Haifeng Wang. 2021. https://doi.org/10.18653/v1/2021.naacl-main.466 R ocket QA : An optimized training approach to dense passage retrieval for open-domain question answering . In ...

  20. [28]

    Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, Ilya Sutskever, et al. 2019. Language models are unsupervised multitask learners. OpenAI blog, 1(8):9

  21. [29]

    Leonardo FR Ribeiro, Mohit Bansal, and Markus Dreyer. 2023. Generating summaries with controllable readability levels. arXiv preprint arXiv:2310.10623

  22. [30]

    Keshav Santhanam, Omar Khattab, Jon Saad-Falcon, Christopher Potts, and Matei Zaharia. 2022. https://doi.org/10.18653/v1/2022.naacl-main.272 C ol BERT v2: Effective and efficient retrieval via lightweight late interaction . In Proceedings of the 2022 Conference of the North Am...

  23. [31]

    Nandan Thakur, Nils Reimers, Andreas R \"u ckl \'e , Abhishek Srivastava, and Iryna Gurevych. 2021. Beir: A heterogeneous benchmark for zero-shot evaluation of information retrieval models. In Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchm...

  24. [32]

    Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, et al. 2023. Llama 2: Open foundation and fine-tuned chat models. arXiv preprint arXiv:2307.09288

  25. [33]

    Bo-Hsiang Tseng, Sheng-Syun Shen, Hung-Yi Lee, and Lin-Shan Lee. 2016. Towards machine comprehension of spoken content: Initial toefl listening comprehension test by machine. In INTERSPEECH

  26. [34]

    Jianyou Wang, Kaicheng Wang, Xiaoyue Wang, Prudhviraj Naidu, Leon Bergen, and Ramamohan Paturi. 2023. https://doi.org/10.48550/arXiv.2310.04678 Doris-mae: Scientific document retrieval using multi-level aspect-based queries . CoRR, abs/2310.04678

  27. [35]

    Wenhui Wang, Furu Wei, Li Dong, Hangbo Bao, Nan Yang, and Ming Zhou. 2020. Minilm: Deep self-attention distillation for task-agnostic compression of pre-trained transformers. Advances in Neural Information Processing Systems, 33:5776--5788

  28. [36]

    Yining Wang, Liwei Wang, Yuanzhi Li, Di He, and Tie-Yan Liu. 2013. A theoretical analysis of ndcg type ranking measures. In Conference on learning theory, pages 25--54. PMLR

  29. [37]

    Lee Xiong, Chenyan Xiong, Ye Li, Kwok-Fung Tang, Jialin Liu, Paul N Bennett, Junaid Ahmed, and Arnold Overwijk. 2020. Approximate nearest neighbor negative contrastive learning for dense text retrieval. In International Conference on Learning Representations

  30. [38]

    Jingtao Zhan, Jiaxin Mao, Yiqun Liu, Jiafeng Guo, Min Zhang, and Shaoping Ma. 2021. Optimizing dense retrieval model training with hard negatives. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, pages 1503--1512

  31. [39]

    Hang Zhang, Yeyun Gong, Yelong Shen, Jiancheng Lv, Nan Duan, and Weizhu Chen. 2021. Adversarial retriever-ranker for dense text retrieval. In International Conference on Learning Representations

  32. [40]

    Tianyi Zhang, Faisal Ladhak, Esin Durmus, Percy Liang, Kathleen McKeown, and Tatsunori B Hashimoto. 2024. Benchmarking large language models for news summarization. Transactions of the Association for Computational Linguistics, 12:39--57

  33. [41]

    Yucheng Zhou, Tao Shen, Xiubo Geng, Chongyang Tao, Can Xu, Guodong Long, Binxing Jiao, and Daxin Jiang. 2023. https://doi.org/10.18653/v1/2023.findings-acl.332 Towards robust ranker for text retrieval . In Findings of the Association for Computational Linguistics: ACL 2023, pa...

  34. [42]

    Hanlin Zhu, Andrew Cohen, Danqing Wang, Kevin Yang, Xiaomeng Yang, Jiantao Jiao, and Yuandong Tian. 2023. https://arxiv.org/abs/2310.08796 End-to-end story plot generator . Preprint, arXiv:2310.08796

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.