Pith. sign in

REVIEW 3 major objections 5 minor 62 references

Hierarchical Level-Wise News Article Clustering via Multilingual Matryoshka Embeddings

T0 review · 3 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 Matryoshka embeddings sort news by story, topic, and theme in one vector.

desk verdict A solid engineering contribution undermined by a potential train/test contamination that the paper never rules out; the abstract also overclaims the SOTA by dropping the bi-encoder qualifier. read the letter →

arxiv 2506.00277 v1 pith:EULQHFXN submitted 2025-05-30 cs.CL cs.AIcs.SI

Hans W. A. Hanley , Zakir Durumeric This is my paper
classification cs.CLcs.AIcs.SI
keywords MatryoshkaembeddingshierarchicalnewsclusteringmultilingualsimilarityreciprocalagglomerativetopicmodelingSemEval2022Task8interpretablenarrativetracking
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

News articles rarely cluster at one clean level: the same event, its broader topic, and its overarching theme are different groupings. The paper argues that Matryoshka embeddings, vectors whose early dimensions encode coarse meaning and later dimensions encode fine detail, can capture this hierarchy directly so that a single multilingual encoder can tell whether two articles are the same story, the same topic, or merely the same theme by examining different slices of the same vector. On the SemEval-2022 Task 8 test set the trained model reaches a Pearson correlation of 0.816, above the previous best bi-encoder. The paper's new level-wise agglomerative clustering then uses these slices to build a three-level tree and reports better clustering F1 than BERTopic at every granularity on the same benchmark. If this holds, downstream work on media monitoring, narrative tracking, and misinformation analysis can replace opaque single-scale topic models with a cheap, interpretable, multilingual hierarchy.

What carries the argument

The central object is a Matryoshka embedding, a contextual vector trained so that nested prefixes of its dimensions capture increasingly fine-grained news similarity. The paper modifies the AngIE contrastive loss, a ranking loss that compares cosine similarities and complex-plane angle differences, to treat different SemEval similarity grades as positives at different prefix lengths, and adds SimCSE dropout positives for monolingual stability. The companion mechanism is Level-Wise RAC: a reciprocal-nearest-neighbor agglomerative clustering algorithm that runs three merge passes, using the d/4 slice for themes, the d/2 slice for topics, and the full vector for stories, with thresholds lambda_1, lambda_2, lambda_3 fixed on validation.

What would settle it

Give the pipeline a new multilingual news corpus with independent, human-annotated story, topic, and theme labels that are not derived from SemEval grades, then compare F1 at each level. If the Level-Wise RAC thresholds fixed from SemEval validation do not recover those annotations, especially at the topic level, the claimed hierarchy is a property of the training labels rather than a general structure of news.

Watch

Extended reading notes

Core claim

Starting from the SemEval-2022 Task 8 similarity labels, the authors train a multilingual Matryoshka embedding by applying a modified AngIE contrastive loss at three nested dimensionality levels: at d/4 dimensions only "Very Similar" pairs count as similar, at d/2 dimensions "Somewhat Similar" and above count, and at the full d dimensions all non-dissimilar pairs count. This encoding forces the leading dimensions to represent broad thematic agreement and later dimensions to represent event-level detail. Building on that, the paper introduces Level-Wise RAC, a reciprocal-nearest-neighbor agglomerative clustering that merges clusters at the coarse dimension slice, then the middle slice, then the full vector, with merge thresholds chosen on validation to maximize F1. The result, the authors claim, is a state-of-the-art multilingual news embedding (Pearson rho = 0.816 on the extended SemEval test set) and a clustering method that identifies stories, topics, and themes at F1 = 0.849 at the story level, outperforming BERTopic at each level of granularity.

Load-bearing premise

The paper assumes that the four SemEval human similarity grades ("Very Similar," "Somewhat Similar," "Somewhat Dissimilar," "Very Dissimilar") line up cleanly with the three hierarchy levels the method outputs, and that merge thresholds tuned on that dataset's validation split will transfer to new news data.

Editorial extensions

If this is right

  • One multilingual encoder replaces separate story, topic, and theme models; the same 768-dimension vector is sliced at 192, 384, and 768 dimensions to answer three similarity questions.
  • News monitoring systems can track a single event across languages and writing styles without pre-specifying the number of clusters, since Level-Wise RAC decides cluster boundaries from validation-tuned similarity thresholds.
  • The hierarchy is interpretable: cluster summaries and class-based TF-IDF keywords give human-readable English story names, topics, and themes for multilingual clusters.
  • The embeddings retain high performance on a 54-language extended test set, with relational similarity to English averaging 0.753, so the approach reaches beyond the original ten SemEval languages.

Reading between the lines

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

  • Because the thresholds lambda_1, lambda_2, lambda_3 are tuned to maximize F1 on the same SemEval validation split whose labels define the hierarchy, the story/topic/theme distinction could partly be an artifact of those four human grades; testing on an independent corpus with its own three-level annotation would separate discovered hierarchy from tuning.
  • A natural extension is to replace the fixed prefixes (d/4, d/2, d) with adaptive dimension selection per language or per corpus, which might improve low-resource languages such as Burmese, where the paper reports relational similarity to English of only 0.452.
  • The same nested-embedding trick could be applied to social media posts or long documents, where story/topic/theme structure is noisier; the paper's own case studies suggest the method is aimed at that broader monitoring setting.
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

3 major / 5 minor

Summary. The paper introduces a multilingual Matryoshka embedding model and a level-wise hierarchical clustering algorithm (Level-Wise RAC) for news article clustering. The embeddings are trained with a modified AngIE loss applied at three nested dimension cuts (d/4, d/2, d), where lower dimensions are trained to separate broad similarity classes and higher dimensions to separate finer ones. The model is trained on an augmented version of SemEval-2022 Task 8, expanded via GPT-4o rewrites, translations, and SimCSE-style self-pairs to 4.10M pairs. The authors report state-of-the-art Pearson correlation on the SE-22-t8 test set for the fine-tuned mE5-base model (0.817 vs. 0.801 for GateNLP-UShef) and show that the Matryoshka model achieves an F1 of 0.849 for story-level clustering with their RAC method, outperforming BERTopic at all granularities. The paper includes ablations, cross-lingual isomorphism measurements, and qualitative case studies on real-world news datasets.

Significance. If the reported numbers hold, the paper makes a useful contribution: it shows that Matryoshka representations can encode not just budget-friendly embeddings but also task-relevant semantic granularity, and that this structure can be exploited by a modified RAC algorithm for interpretable hierarchical clustering. The multilingual extension via GPT-4o translations is practical, and the authors release model weights and synthetic portions of the data, which supports reproducibility. The main scientific claims are the SOTA similarity score and the hierarchical clustering F1 improvements over BERTopic; both are contingent on an unresolved train/test separation issue. The paper also demonstrates careful ablations (SimCSE positives, data augmentation) and provides a concrete evaluation pipeline, which are strengths.

major comments (3)
  1. [Section 3, Dataset (paras. 1 and 4)] The manuscript never states that the SE-22-t8 test split was excluded from the augmented training pool. Section 3 reports that all 37,394 SE-22-t8 URLs were scraped, that the resulting articles were rewritten/translated and combined into 4.10M pairs, and that 10% was held out as validation. The test split of 3,958 pairs and 7,842 unique articles is introduced only in the next paragraph. Because the training set includes SimCSE self-pairs and GPT-4o rewrite/translation pairs derived from the scraped articles, any test article that survived scraping and filtering could appear in training as an unlabeled or labeled positive. This would inflate the Table 1 Pearson correlations (e.g., 0.817 for fine-mE5-base) and the Table 4 clustering F1 scores, and it would make the comparison with GateNLP-UShef, which was trained on the official train split, unfair. The authors must explicitly state whether test articles and test pairs were excluded from all stages of augmentation and training, and ideally report a check for overlap between training and test article IDs or texts.
  2. [Abstract and §5.1, Table 1] The abstract's claim 'state-of-the-art performance on the SemEval 2022 Task 8 test dataset (Pearson ρ = 0.816)' is not what Table 1 shows. The value 0.816 is the Pearson correlation of mat-mE5-base-384 on the extended, GPT-4o-translated test set; on the original 10-language SE-22-t8 test set, the same model scores 0.792. The only model that beats GateNLP-UShef on the original test set is fine-mE5-base at 0.817, which §5.1 correctly qualifies as 'for bi-cross embedding models.' The abstract conflates the original and extended test sets and drops the qualifying comparison class. This should be corrected so the headline result matches the evidence, and the 'state-of-the-art' wording should be scoped to the bi-encoder setting used in the evaluation.
  3. [Section 4 and §5.3, Table 4] The hierarchical clustering evaluation is partly self-confirming. The training objective Lmat (Section 4) uses the SE-22-t8 label tiers to define coarse-to-fine similarity thresholds at dimensions d/4, d/2, and d, and the RAC algorithm's λ thresholds are tuned on the validation split of the same dataset to maximize F1 for those same tiers. Consequently, Table 4's F1 scores measure how well the pipeline reproduces the label hierarchy that was built into training and threshold tuning; they do not independently validate the story/topic/theme structure against an external taxonomy. The paper should either validate the discovered levels on an external hierarchical dataset (e.g., event-coreference or topic-hierarchy benchmarks) or explicitly frame Table 4 as an intrinsic evaluation of the proposed pipeline, not as evidence of discovering ground-truth hierarchical news structure.
minor comments (5)
  1. [Abstract and Section 5.1] The term 'bi-cross embedding models' appears in the abstract and text; the standard term in the literature is 'bi-encoder.' Please clarify the intended meaning or use the standard terminology.
  2. [Introduction, first paragraph] Typo: 'do no not scale' should be 'do not scale.' In the same paragraph, 'genetic language models' should be 'generative language models' (this typo also appears in the Entity Sensitivity paragraph).
  3. [Section 5.3, Table 3] The text says the Matryoshka model 'perform[s] largely the best' on the 20 NewsGroup dataset, but in the BERTopic column fine-mE5-base achieves F1 = 0.2738, slightly higher than mat-mE5-base-192's 0.2730. Please adjust the wording or analysis to match the table.
  4. [Section 5.2 and Table 1] The fine-tuned umt5-base model drops from 0.815 on the original test set to 0.582 on the extended set, a large and unexplained degradation. A sentence of explanation (or caveat) would help readers interpret the comparison.
  5. [Figure 2 caption and labels] The caption says 'utilizing Matryoshka e5-base embeddings' but panels (a)–(d) show non-Matryoshka fine-tuned models; the model names in the subcaptions also contain spacing errors (e.g., 'matryoshka-mE5-d/4dimensions'). Please make the figure labels and caption consistent.

Circularity Check

2 steps flagged · score 6.0 of 10

The reported test-set state of the art is compromised because the 4.10M-pair augmented training pool is built from all scraped URLs with no stated exclusion of the SE-22-t8 test articles, and the story/topic/theme hierarchy is defined by the same label thresholds used to supervise the model.

  1. fitted input called prediction [Section 3 (Dataset), 'Dataset Augmentations' and 'Test Dataset' paragraphs]
    "we scraped each of the 37,394 URLs provided. After removing dead links... we were left with 24,871 articles... Altogether, after rewriting and translating the original set of articles... we extend the original dataset to 4.10M article pairs. We utilize 10% (410K article pairs) of our SE-22-t8 dataset as validation. ... Test Dataset. For our evaluation, in addition to the original 10-language SE-22-t8 test split of 3,958 article pairs and 7,842 unique articles..."

    The 37,394 scraped URLs include the test split's URLs, and the 4.10M-pair augmentation is built from all scraped articles with no stated exclusion of the test split. The 10% validation is drawn from these augmented pairs, so test-derived article texts and their GPT-4o rewrites/translations can appear in training or validation. The reported test-set Pearson rho = 0.817 and clustering F1 = 0.849 are then measurements on articles whose content was already used to fit the model and select thresholds, so the claimed state-of-the-art prediction on the test split reduces to a fit on data derived from that split.

  2. self definitional [Section 4, 'Modified AngIE Loss for MRL' and Section 5.3, 'Hierarchical Clustering']
    "for the loss applied at d/4-dimensions we treat 'Very Dissimilar' pairs as having a labeled cosine similarity of 0, and all other pairs as having a labeled cosine similarity of 1; then for the loss applied at d/2 dimensions, we treat 'Very Dissimilar' and 'Somewhat Dissimilar' pairs as having a labeled cosine similarity of 0, and all other pairs as having a labeled cosine similarity of 1,etc... This forces the embedding to progressively learn to differentiate different levels of similarity..."

    The hierarchy of story/topic/theme is defined by thresholds on the same OVERALL similarity labels that supervise the Matryoshka loss at d/4, d/2, and d, and the RAC merge thresholds are tuned to maximize F1 on those same label tiers. The clustering evaluation therefore checks whether the model reproduces the label partitions it was trained to produce; the claimed discovery of stories, topics, and themes is a re-instantiation of the training label mapping rather than an independently derived hierarchy.

full rationale

The central derivation chain is not self-contained as written. First, the dataset construction does not hold out the test split before augmentation: all 37,394 URLs are scraped, rewritten, and translated into 4.10M article pairs, and only afterward is the test split described. Because the paper never states that the 7,842 unique test articles were excluded from this augmented pool, the reported test Pearson correlation and clustering F1 may be evaluated on articles whose content already appeared in training or validation pairs. If so, the state-of-the-art claim on SE-22-t8 is statistically forced rather than predictive. Second, even setting leakage aside, the hierarchical levels are self-definitional: the Matryoshka loss thresholds and the RAC λ thresholds are both derived from the same OVERALL label tiers, so the clustering F1 at story/topic/theme granularity measures the model's ability to reproduce the label mapping it was trained on. The paper does provide some independent evidence, including external clustering benchmarks on Miranda et al. (2018) and 20 NewsGroup, and the embedding model is compared against a held-out human-labeled test split if leakage is absent. These factors make the circularity partial rather than total, but the missing test-set exclusion and the label-defined hierarchy prevent the central claims from being fully independent of their inputs.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on a handful of design choices: the label-to-level mapping, the dimension cut points, the RAC thresholds, and the assumption that GPT-4o synthetic translations preserve similarity. These are explicit modeling choices rather than hidden entities.

free parameters (3)
  • RAC merge thresholds lambda_1, lambda_2, lambda_3 = not reported
    Chosen on SE-22-t8 validation to maximize F1; values not disclosed, and clustering results depend on them.
  • Dimension cut points (d/4, d/2, d) = d/4, d/2, d
    Hand-chosen boundaries for theme/topic/story levels; not optimized or justified by data.
  • Similarity label mapping per level = Very Similar, Somewhat Similar, Somewhat Dissimilar
    The mapping from four similarity categories to binary positives at each dimension is a modeling choice that defines the hierarchy.
assumptions (4)
  • domain assumption SemEval OVERALL similarity categories map to a story/topic/theme hierarchy
    Assumed in Sections 3 and 4; no external grounding that Very Similar equals same story, Somewhat Similar equals same topic, and so on.
  • domain assumption GPT-4o translations and rewrites preserve semantic similarity and story identity
    Used to extend training and test sets in Section 3; systematic translation errors would bias both training and evaluation.
  • domain assumption MRL training makes higher dimensions encode finer-grained similarity
    The paper relies on this property (Figure 1) after imposing it via loss thresholds; this is not proven independently of the training objective.
  • standard math RAC reciprocal nearest neighbor merging yields valid clusters for text data
    Adopted from Sumengen et al. 2021 as the base algorithm; treated as a standard tool.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Hierarchical Level-Wise News Article Clustering via Multilingual Matryoshka Embeddings." pith.science (2026). https://pith.science/paper/EULQHFXN

@misc{pith2026250600277,
  author       = {Pith},
  title        = {Pith review of: Hierarchical Level-Wise News Article Clustering via Multilingual Matryoshka Embeddings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EULQHFXN}},
  note         = {Machine review of arXiv:2506.00277}
}
abstract

Contextual large language model embeddings are increasingly utilized for topic modeling and clustering. However, current methods often scale poorly, rely on opaque similarity metrics, and struggle in multilingual settings. In this work, we present a novel, scalable, interpretable, hierarchical, and multilingual approach to clustering news articles and social media data. To do this, we first train multilingual Matryoshka embeddings that can determine story similarity at varying levels of granularity based on which subset of the dimensions of the embeddings is examined. This embedding model achieves state-of-the-art performance on the SemEval 2022 Task 8 test dataset (Pearson $\rho$ = 0.816). Once trained, we develop an efficient hierarchical clustering algorithm that leverages the hierarchical nature of Matryoshka embeddings to identify unique news stories, narratives, and themes. We conclude by illustrating how our approach can identify and cluster stories, narratives, and overarching themes within real-world news datasets.

Figures

Figures reproduced from arXiv: 2506.00277 by the authors.

Figure 1
Figure 1. Matryoshka Representation Learning for em [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Cosine similarity of test split of the SE-22-t8 dataset article pair embeddings utilizing Matryoshka e5-base [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Example summary of multilingual documents. [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

62 extracted references · 44 canonical work pages

  1. [1]

    Aly Abdelrazek, Yomna Eid, Eman Gawish, Walaa Medhat, and Ahmed Hassan. 2023. Topic modeling algorithms and applications: A survey. Information Systems, 112:102131

    work page 2023

  2. [2]

    Mohiuddin Ahmed, Raihan Seraj, and Syed Mohammed Shamsul Islam. 2020. The k-means algorithm: A comprehensive survey and performance evaluation. Electronics, 9(8):1295

    work page 2020

  3. [3]

    MohammadHossein Bateni, Laxman Dhulipala, Kishen N Gowda, D Ellis Hershkowitz, Rajesh Jayaram, and Jakub a cki. 2024. It’s hard to hac average linkage! In 51st International Colloquium on Automata, Languages, and Programming (ICALP 2024), pages 18--1. Schloss Dagstuhl--Leibniz-Zentrum f \"u r Informatik

    work page 2024

  4. [4]

    Desmond Bala Bisandu, Rajesh Prasad, and Musa Muhammad Liman. 2018. Clustering news articles using efficient similarity measure and n-grams. International Journal of Knowledge Engineering and Data Mining, 5(4):333--348

    work page 2018

  5. [5]

    Stephen Blank. 2022. Russia, china, and information war against ukraine. The Journal of East Asian Affairs, pages 39--72

    work page 2022

  6. [6]

    Daniel Cer, Yinfei Yang, Sheng-yi Kong, Nan Hua, Nicole Limtiaco, Rhomni St John, Noah Constant, Mario Guajardo-Cespedes, Steve Yuan, Chris Tar, et al. 2018. Universal sentence encoder for english. In Proceedings of the 2018 conference on empirical methods in natural language processing: system demonstrations, pages 169--174

    work page 2018

  7. [7]

    Nathanael Chambers and Dan Jurafsky. 2009. Unsupervised learning of narrative schemas and their participants. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the 4th International Joint Conference on Natural Language Processing of the AFNLP, pages 602--610

    work page 2009

  8. [8]

    Changyou Chen, Jianyi Zhang, Yi Xu, Liqun Chen, Jiali Duan, Yiran Chen, Son Tran, Belinda Zeng, and Trishul Chilimbi. 2022 a . Why do we need large batchsizes in contrastive learning? a gradient-bias perspective. Advances in Neural Information Processing Systems, 35:33860--33875

    work page 2022

Show all 62 references
  1. [9]

    Xi Chen, Mattia Samory, Scott Hale, David Jurgens, and Przemyslaw A Grabowicz. 2024. A multilingual similarity dataset for news article frame. In Proceedings of the International AAAI Conference on Web and Social Media, volume 18, pages 1913--1923

    2024

  2. [10]

    Xi Chen, Ali Zeynali, Chico Camargo, Fabian Fl \"o ck, Devin Gaffney, Przemyslaw Grabowicz, Scott Hale, David Jurgens, and Mattia Samory. 2022 b . Semeval-2022 task 8: Multilingual news article similarity. In Proceedings of the 16th International Workshop on Semantic Evaluatio...

    2022

  3. [11]

    Or Dinari and Oren Freifeld. 2022. Revisiting dp-means: fast scalable algorithms via parallelism and delayed cluster creation. In Uncertainty in Artificial Intelligence, pages 579--588. PMLR

    2022

  4. [12]

    Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783

    2024 arXiv

  5. [13]

    Alexander Richard Fabbri, Irene Li, Tianwei She, Suyi Li, and Dragomir Radev. 2019. Multi-news: A large-scale multi-document summarization dataset and abstractive hierarchical model. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pa...

    2019

  6. [14]

    Max Falkenberg, Alessandro Galeazzi, Maddalena Torricelli, Niccol \`o Di Marco, Francesca Larosa, Madalina Sas, Amin Mekacher, Warren Pearce, Fabiana Zollo, Walter Quattrociocchi, et al. 2022. Growing polarization around climate change on social media. Nature Climate Change, 1...

    2022

  7. [15]

    Saumya Gandhi, Ritu Gala, Vijay Viswanathan, Tongshuang Wu, and Graham Neubig. 2024. Better synthetic data by retrieving and transforming existing datasets. In Findings of the Association for Computational Linguistics ACL 2024, pages 6453--6466

    2024

  8. [16]

    T Gao, X Yao, and Danqi Chen. 2021. Simcse: Simple contrastive learning of sentence embeddings. In EMNLP 2021-2021 Conference on Empirical Methods in Natural Language Processing, Proceedings

    2021

  9. [17]

    Yunfan Gao, Yun Xiong, Xinyu Gao, Kangxiang Jia, Jinliu Pan, Yuxi Bi, Yi Dai, Jiawei Sun, and Haofen Wang. 2023. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997

    2023 arXiv

  10. [18]

    Maarten Grootendorst. 2022. Bertopic: Neural topic modeling with a class-based tf-idf procedure. arXiv preprint arXiv:2203.05794

    2022 arXiv

  11. [19]

    Hans WA Hanley and Zakir Durumeric. 2024. Machine-made media: Monitoring the mobilization of machine-generated articles on misinformation and mainstream news websites. In Proceedings of the International AAAI Conference on Web and Social Media, volume 18, pages 542--556

    2024

  12. [20]

    Hans WA Hanley, Deepak Kumar, and Zakir Durumeric. 2023. Happenstance: Utilizing semantic search to track russian state media narratives about the russo-ukrainian war on reddit. In Proceedings of the international AAAI conference on web and social media, volume 17, pages 327--338

    2023

  13. [21]

    Hans WA Hanley, Deepak Kumar, and Zakir Durumeric. 2024. Specious sites: Tracking the spread and sway of spurious news stories at scale. In 2024 IEEE Symposium on Security and Privacy (SP), pages 180--180. IEEE Computer Society

    2024

  14. [22]

    Hans WA Hanley, Emily Okabe, and Zakir Durumeric. 2025. Tracking the takes and trajectories of english-language news narratives across trustworthy and worrisome websites. 34th USENIX Security Symposium (USENIX Security 25)

    2025

  15. [23]

    Edward J Hu, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, Weizhu Chen, et al. 2021. Lora: Low-rank adaptation of large language models. In International Conference on Learning Representations

    2021

  16. [24]

    Aaron Hurst, Adam Lerer, Adam P Goucher, Adam Perelman, Aditya Ramesh, Aidan Clark, AJ Ostrow, Akila Welihinda, Alan Hayes, Alec Radford, et al. 2024. Gpt-4o system card. arXiv preprint arXiv:2410.21276

    2024 arXiv

  17. [25]

    Hamed Jelodar, Yongli Wang, Chi Yuan, Xia Feng, Xiahui Jiang, Yanchao Li, and Liang Zhao. 2019. Latent dirichlet allocation (lda) and topic modeling: models, applications, a survey. Multimedia tools and applications, 78:15169--15211

    2019

  18. [26]

    Ke Jiang, Brian Kulis, and Michael Jordan. 2012. Small-variance asymptotics for exponential family dirichlet process mixture models. Advances in Neural Information Processing Systems, 25

    2012

  19. [27]

    Diederik P Kingma and Jimmy Lei Ba. 2015. Adam: A method for stochastic gradient descent. In ICLR: international conference on learning representations, pages 1--15. ICLR US

    2015

  20. [28]

    Aditya Kusupati, Gantavya Bhatt, Aniket Rege, Matthew Wallingford, Aditya Sinha, Vivek Ramanujan, William Howard-Snyder, Kaifeng Chen, Sham Kakade, Prateek Jain, et al. 2022. Matryoshka representation learning. Advances in Neural Information Processing Systems, 35:30233--30249

    2022

  21. [29]

    Guillaume Lample, Alexis Conneau, Marc'Aurelio Ranzato, Ludovic Denoyer, and Herv \'e J \'e gou. 2018. Word translation without parallel data. In International conference on learning representations

    2018

  22. [30]

    Matthew Le, Stephen Roller, Laetitia Papaxanthos, Douwe Kiela, and Maximilian Nickel. 2019. Inferring concept hierarchies from text corpora via hyperbolic embeddings. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 3231--3241

    2019

  23. [31]

    Chong Li, Shaonan Wang, Jiajun Zhang, and Chengqing Zong. 2024 a . Improving in-context learning of multilingual generative language models with cross-lingual alignment. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Li...

    2024

  24. [32]

    Xianming Li and Jing Li. 2024. Aoe: Angle-optimized embeddings for semantic textual similarity. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1825--1839

    2024

  25. [33]

    Xianming Li, Zongxi Li, Jing Li, Haoran Xie, and Qing Li. 2024 b . 2d matryoshka sentence embeddings. arXiv preprint arXiv:2402.14776

    2024 arXiv

  26. [34]

    Synthetic data generation with large language models for text classification: Potential and limitations

    Zhuoyan Li, Hangxiao Zhu, Zhuoran Lu, and Ming Yin. Synthetic data generation with large language models for text classification: Potential and limitations. In The 2023 Conference on Empirical Methods in Natural Language Processing

    2023

  27. [35]

    Kelly Marchisio, Neha Verma, Kevin Duh, and Philipp Koehn. 2022. Isovec: Controlling the relative isomorphism of word embedding spaces. In Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pages 6019--6033

    2022

  28. [36]

    Leland McInnes, John Healy, Steve Astels, et al. 2017. hdbscan: Hierarchical density based clustering. J. Open Source Softw., 2(11):205

    2017

  29. [37]

    Leland McInnes, John Healy, Nathaniel Saul, and Lukas Gro berger. 2018. Umap: Uniform manifold approximation and projection. Journal of Open Source Software, 3(29)

    2018

  30. [38]

    Yu Meng, Yunyi Zhang, Jiaxin Huang, Yu Zhang, Chao Zhang, and Jiawei Han. 2020. Hierarchical topic mining via joint spherical tree and text embedding. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, pages 1908--1917

    2020

  31. [39]

    Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. 2013. Distributed representations of words and phrases and their compositionality. Advances in neural information processing systems, 26

    2013

  32. [40]

    Sebasti \ a o Miranda, Arturs Znotins, Shay B Cohen, and Guntis Barzdins. 2018. Multilingual clustering of streaming news. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 4535--4544

    2018

  33. [41]

    Eric Mitchell, Yoonho Lee, Alexander Khazatsky, Christopher D Manning, and Chelsea Finn. 2023. Detectgpt: zero-shot machine-generated text detection using probability curvature. In Proceedings of the 40th International Conference on Machine Learning, pages 24950--24962

    2023

  34. [42]

    Nicholas Monath, Kumar Avinava Dubey, Guru Guruganesh, Manzil Zaheer, Amr Ahmed, Andrew McCallum, Gokhan Mergen, Marc Najork, Mert Terzihan, Bryon Tjanaka, et al. 2021. Scalable hierarchical agglomerative clustering. In Proceedings of the 27th ACM SIGKDD Conference on knowledg...

    2021

  35. [43]

    Nicholas Monath, Manzil Zaheer, and Andrew McCallum. 2023. Online level-wise hierarchical clustering. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1733--1745

    2023

  36. [44]

    Dan S Nielsen and Ryan McConville. 2022. Mumin: A large-scale multilingual multimodal fact-checked misinformation social network dataset. In Proceedings of the 45th international ACM SIGIR conference on research and development in information retrieval, pages 3141--3153

    2022

  37. [45]

    Minsu Park, Seyeon Choi, Chanyeol Choi, Jun-Seong Kim, and Jy-Yong Sohn. 2024. Improving multi-lingual alignment through soft contrastive learning. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Langu...

    2024

  38. [46]

    Jeffrey Pennington, Richard Socher, and Christopher D Manning. 2014. Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP), pages 1532--1543

    2014

  39. [47]

    Juan Ramos et al. 2003. Using tf-idf to determine word relevance in document queries. In Proceedings of the first instructional conference on machine learning, volume 242, pages 29--48. Citeseer

    2003

  40. [48]

    Nils Reimers and Iryna Gurevych. 2019. Sentence-bert: Sentence embeddings using siamese bert-networks. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJC...

    2019

  41. [49]

    Jan Rupnik, Andrej Muhic, Gregor Leban, Primoz Skraba, Blaz Fortuna, and Marko Grobelnik. 2016. News across languages-cross-lingual document similarity and event tracking. Journal of Artificial Intelligence Research, 55:283--316

    2016

  42. [50]

    Gregory W Schwartz, Yeqiao Zhou, Jelena Petrovic, Maria Fasolino, Lanwei Xu, Sydney M Shaffer, Warren S Pear, Golnaz Vahedi, and Robert B Faryabi. 2020. Toomanycells identifies and visualizes relationships of single-cell clades. Nature methods, 17(4):405--413

    2020

  43. [51]

    Iknoor Singh, Yue Li, Melissa Thong, and Carolina Scarton. 2022. Gatenlp-ushef at semeval-2022 task 8: Entity-enriched siamese transformer for multilingual news article similarity. In Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pages 1...

    2022

  44. [52]

    Roger Alan Stein, Patricia A Jaques, and Joao Francisco Valiati. 2019. An analysis of hierarchical text classification using word embeddings. Information Sciences, 471:216--232

    2019

  45. [53]

    Baris Sumengen, Anand Rajagopalan, Gui Citovsky, David Simcha, Olivier Bachem, Pradipta Mitra, Sam Blasiak, Mason Liang, and Sanjiv Kumar. 2021. Scaling hierarchical agglomerative clustering to billion-sized datasets. arXiv preprint arXiv:2105.11653

    2021 arXiv

  46. [54]

    Zhiqing Sun, Zhi-Hong Deng, Jian-Yun Nie, and Jian Tang. 2019. Rotate: Knowledge graph embedding by relational rotation in complex space. In International Conference on Learning Representations

    2019

  47. [55]

    Liang Wang, Nan Yang, Xiaolong Huang, Binxing Jiao, Linjun Yang, Daxin Jiang, Rangan Majumder, and Furu Wei. 2022. Text embeddings by weakly-supervised contrastive pre-training. arXiv preprint arXiv:2212.03533

    2022 arXiv

  48. [56]

    Liang Wang, Nan Yang, Xiaolong Huang, Linjun Yang, Rangan Majumder, and Furu Wei. 2024. Multilingual e5 text embeddings: A technical report. arXiv preprint arXiv:2402.05672

    2024 arXiv

  49. [57]

    Zihang Xu, Ziqing Yang, Yiming Cui, and Zhigang Chen. 2022. Hfl at semeval-2022 task 8: A linguistics-inspired regression model with data augmentation for multilingual news similarity. In Proceedings of the 16th International Workshop on Semantic Evaluation (SemEval-2022), pag...

    2022

  50. [58]

    Hui Zhang, Tu Bao Ho, Yang Zhang, and M-S Lin. 2006. Unsupervised feature extraction for time series clustering using orthogonal wavelet transform. Informatica, 30(3)

    2006

  51. [59]

    Mozhi Zhang, Keyulu Xu, Ken-ichi Kawarabayashi, Stefanie Jegelka, and Jordan Boyd-Graber. 2019. Are girls neko or sh \=o jo? cross-lingual alignment of non-isomorphic embeddings with iterative normalization. In Proceedings of the 57th Annual Meeting of the Association for Comp...

    2019

  52. [60]

    Xin Zhang, Yanzhao Zhang, Dingkun Long, Wen Xie, Ziqi Dai, Jialong Tang, Huan Lin, Baosong Yang, Pengjun Xie, Fei Huang, et al. 2024. mgte: Generalized long-context text representation and reranking models for multilingual text retrieval. In EMNLP (Industry Track)

    2024

  53. [61]

    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...

  54. [62]

    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 gl...

Pith tools

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