Ranking-based Fusion Algorithms for Extreme Multi-label Text Classification (XMTC)

Celso Fran\c{c}a; Gestefane Rabbi; Leonardo Rocha; Marcos Andr\'e Gon\c{c}alves; Thiago Salles; Washington Cunha

REVIEW 4 major objections 4 minor 21 references

Ranking-based Fusion Algorithms for Extreme Multi-label Text Classification (XMTC)

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Ranking fusion with ZMUV normalization and CombMNZ is the best-performing combination for fusing sparse and dense retrievers in extreme multi-label text classification.

desk verdict A wide but flawed fusion benchmark for XMTC: the central claim is contradicted by the paper's own tables, and missing single-retriever baselines make the core premise untestable. read the letter →

arxiv 2507.03761 v1 pith:7FZQQ6HP submitted 2025-07-04 cs.IR

Celso França , Gestefane Rabbi , Thiago Salles , Washington Cunha , Leonardo Rocha , Marcos André Gonçalves This is my paper · ORCID

classification cs.IR

keywords ExtremeMulti-labelTextClassificationrankingfusionnormalizationCombMNZZMUVBM25BERTlong-taillabels

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

Extreme multi-label text classification asks a system to rank each document against a huge label vocabulary, where most labels are rare. This paper asks whether fusing the rankings of two very different retrievers—BM25, which matches words exactly, and a fine-tuned BERT model, which matches by semantic similarity—can improve that ranking, and which fusion recipe does it best. After testing six ways of normalizing scores and ten fusion algorithms on four benchmark datasets, it claims the winner is ZMUV normalization (converting each retriever's scores to zero mean and unit variance) followed by CombMNZ, which sums normalized scores and boosts labels that both retrievers support. The reported result is that this combination gives the highest ranking-quality scores at positions 1, 5 and 10, for both head and tail labels on every dataset. If correct, it is a drop-in improvement: no retraining or architecture change, just a pre-processing step on the retrievers' outputs.

What carries the argument

The load-bearing mechanism is the pair (ZMUV, CombMNZ). ZMUV normalization takes each retriever's raw scores, subtracts the mean and divides by the standard deviation, putting BM25's unbounded scores and BERT's similarity scores on one comparable scale. CombMNZ then computes, for each label, the sum of its normalized scores and multiplies by the number of retrievers with a non-zero score for that label, so a label supported by both retrievers is amplified while a label seen by only one is discounted. The paper's experiments vary six normalizations (Min-Max, Max, Sum, ZMUV, Rank, Borda) and ten fusion algorithms (CombMIN, CombMAX, CombMED, CombSUM, CombANZ, CombMNZ, ISR, Log-ISR, BordaFuse, Condorcet), and this pair is what carries the claimed result.

What would settle it

Take any one of the four datasets and compute the overlap (or Kendall tau) between the BM25 and BERT candidate rankings; if the top-10 lists are nearly identical, then ZMUV+CombMNZ cannot beat either retriever alone, and the claimed fusion advantage would disappear.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that fusing dense and sparse rankings for XMTC works best when the two score lists are first standardized to zero mean and unit variance and then merged with CombMNZ: for each candidate label the normalized scores from BM25 and BERT are summed, and that sum is multiplied by the number of retrievers that gave the label a non-zero score. This consensus-boosting step is what yields the highest nDCG@k and Precision@k on Eurlex-4K, Wiki10-31K, Amazon-670K and AmazonCat-13K, for both the 80% least frequent tail labels and the 20% most frequent head labels, with $k \in \{1,5,10\}$. The claim is explicitly a comparative one: of the 60 normalization-plus-fusion combinations tested, ZMUV+CombMNZ is the top performer on all four datasets.

Load-bearing premise

The load-bearing premise is that BM25 and BERT produce complementary label rankings; the paper asserts this but never measures it, and if the two rankers largely agree, fusion adds nothing and the central claim collapses.

Editorial extensions

If this is right

A system that already has BM25 and BERT candidate lists can adopt ZMUV+CombMNZ as a drop-in post-processing step and, according to the paper, obtain the best measured nDCG@k and Precision@k among the 60 compared combinations.
The winning combination improves both tail and head labels together, so there is no measured tradeoff between rare-label coverage and common-label accuracy.
The recommendation generalizes across dataset scales, from Eurlex-4K with about 4K labels to Amazon-670K with over 670K labels, and across document counts from tens of thousands to millions.
Normalization choice is a first-order decision: the same fusion algorithm can differ by several points in nDCG depending on which normalization it is paired with.

Reading between the lines

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

Editorial inference: ZMUV probably wins because it puts BM25's unbounded positive scores and BERT's similarity scores on a common scale; if that is the mechanism, other distribution-equalizing normalizations (e.g., quantile or rank-based z-scores) should produce similar gains.
Editorial inference: the head-tail split is fixed by a Pareto 80/20 rule; the reported tail improvements might shift if the threshold moved, so the recommendation's robustness to that threshold is a testable open question.
Editorial inference: because the recipe only consumes normalized score lists, it should transfer to other retriever pairs and to ranking tasks beyond XMTC; a direct test on those settings would show whether the mechanism is general.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Two things to know. The paper is a benchmark of standard rank-fusion algorithms on top of BM25 and BERT rankings for XMTC, and as a survey-by-benchmark it has decent coverage: six normalizations, ten fusion algorithms, four datasets, 5-fold CV, standard deviations. That part is honest work. But the paper's own conclusion, that CombMNZ with ZMUV normalization is best, is not supported by the numbers in its tables, and the missing single-retriever baselines mean the core premise that fusion helps is never actually tested.

What is new: applying these well-known metasearch techniques to XMTC with a dense retriever is a routine extension, not an innovation. The specific comparison matrix is not in the literature, but the components are all from Fox and Shaw, Aslam and Montague, and Montague and Aslam. So novelty is low, but the empirical question is sensible.

The soft spots are serious. First, no BM25-only or BERT-only results appear anywhere. The reader cannot tell whether fusion improves on the better retriever alone; the complementarity claim in Section 1 is asserted, never measured. Second, the Section 6 statement is contradicted by the tables: on Eurlex-4k, ZMUV CombSUM has tail nDCG@1 51.8 versus 51.5 for CombMNZ, and head 82.3 versus 81.8; on Wiki10-31k, CombSUM again leads tail nDCG@1 49.5 versus 49.1. So 'highest effectiveness across all datasets' is not true on its face. Third, the t-tests promised in Section 4 are missing. Fourth, the Borda Norm rows are identical across multiple algorithms, including ISR, Log_ISR, BordaFuse, and Condorcet, which strongly suggests a bug in the fusion code or normalization. The paper says source code is available but provides no link or repository.

Who should read this? Practitioners working on XMTC who want a cheap way to boost ranking quality by fusing a sparse and a dense retriever. The question is worth answering, and a revised version with baselines, corrected claims, significance tests, and code could be a useful benchmark. As submitted, I would not send it to peer review; the evidence needed to support the central claim is absent, and the apparent table bug makes the results uncheckable. I'd tell the authors to fix these issues before resubmitting.

Referee Report

4 major / 4 minor

Summary. The paper studies ranking-based fusion algorithms for extreme multi-label text classification (XMTC), combining a BM25 sparse retriever with a fine-tuned BERT dense retriever. Six normalization strategies and ten fusion algorithms are evaluated on four XMTC datasets under head/tail label splits, with nDCG@k and Precision@k as metrics. The central claim, stated in Section 6, is that CombMNZ with ZMUV normalization yields the highest effectiveness across all datasets. The manuscript also asserts that sparse and dense rankings are complementary and that fusion improves label ranking, and it announces statistical significance testing and public code/data release.

Significance. If the central claim were supported, the paper would offer a simple, architecture-agnostic recipe for improving XMTC ranking by fusing BM25 and BERT outputs, which could be of practical value. The evaluation grid is broad (six normalizations, ten fusion algorithms, four datasets, head/tail breakdowns) and the focus on tail labels is relevant to XMTC. However, the paper currently provides no verifiable evidence for its main conclusion: the own tables contradict the 'highest across all datasets' statement, no single-retriever baselines are reported, and the promised significance tests and reproducibility artifacts are absent. The contribution is therefore not established in its present form.

major comments (4)

[Section 6, Tables 2-5] The conclusion that ZMUV+CombMNZ yielded the highest effectiveness across all datasets is contradicted by the paper's own tables. On Eurlex-4k (Table 2), ZMUV+CombSUM achieves tail nDCG@1 of 51.8 vs. 51.5 for ZMUV+CombMNZ and head nDCG@1 of 82.3 vs. 81.8. On Wiki10-31k (Table 3), ZMUV+CombSUM again leads on tail nDCG@1 (49.5 vs. 49.1) and head nDCG@1 (81.8 vs. 81.2). On Amazon-670k (Table 4), ZMUV+CombSUM beats CombMNZ on tail nDCG@1 (48.0 vs. 47.3). The paper never specifies the aggregation rule over the 24 metric cells (2 metrics x 3 cutoffs x head/tail x 4 datasets), so 'highest effectiveness' is undefined. The conclusion must be revised or justified with an explicit ranking criterion.
[Sections 4 and 5] The paper claims that fusion of sparse and dense retrievers improves effectiveness and asserts complementarity (Section 1), but no effectiveness numbers are reported for BM25-only or BERT-only rankings. Every table shows only fused outputs, so the reader cannot determine whether fusion helps at all, nor whether the two retrievers are complementary. Without these single-retriever baselines, the central premise of the paper is untested. The authors should report per-retriever performance and, ideally, a quantitative complementarity measure (e.g., rank correlation between the two input rankings).
[Section 4] The text states that 'we assess the statistical significance of our results by employing the two-sided paired Student's t-test with 95% confidence' and that results are 'averaged across the five test splits', but no t-test outcomes, p-values, or significance markers appear in Tables 2-5. The reported standard deviations alone do not substantiate the significance claim, especially when differences as small as 0.1-0.3 percentage points are discussed. The authors must either report the significance results or remove the claim.
[Section 4] The reproducibility statement says 'we make our source code, developed models, and data publicly available', but no repository URL, DOI, or access instructions are provided anywhere in the manuscript. This makes the empirical results unverifiable, particularly the anomalous 1.3-1.7 nDCG values for Borda Norm with ISR/Log_ISR/BordaFuse/Condorcet in Tables 2-5, which suggest a potential implementation issue (e.g., ties or zero normalization) that cannot be checked without code.

minor comments (4)

[Table 5 caption] The caption of Table 5 reads 'Amazon-13k' while Section 4 and Table 1 refer to 'AmazonCat-13K'; the naming should be made consistent.
[Abstract and Introduction] The abstract and introduction use 'proven effective' for sparse-dense fusion in XMTC, but the paper itself does not compare against the individual retrievers; this overstates what is demonstrated.
[Section 2, ZMUV Norm] The definition of ZMUV Norm states it transforms scores 'into a standard normal distribution'; since the transformation is applied per ranked list, it would be more precise to say it standardizes the scores to zero mean and unit variance for each retrieval run, not that the scores follow a normal distribution.
[References] Several references are incomplete or use placeholder formatting (e.g., the ACM template placeholders for the conference acronym and DOI), and reference [17] (Wei et al., WSDM 2022) appears unrelated to the cyclical learning rate citation in Section 4; this should be checked.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the paper is a self-contained empirical benchmark; its central claim is a comparison of reported results, not a derivation that reduces to its inputs.

full rationale

The paper contains no derivation chain that could reduce to its inputs. Sections 2 and 3 define standard normalization and fusion formulas from the metasearch literature, and Section 4 fixes standard hyperparameters (BM25 b=0.75, k=1.5; BERT fine-tuning with a contrastive objective). The Section 6 conclusion ('Employing the CombMNZ ranking-based fusion algorithm in conjunction with the ZMUV normalization strategy for fusing both dense and sparse rankings yielded the highest effectiveness across all datasets') is a summary of the comparisons in Tables 2-5, not a prediction generated from fitted values; no parameter is fitted on one subset and then reported as a prediction for a closely related quantity. The self-citations [5,6,7,8,15] are used for experimental conventions such as 5-fold cross-validation and benchmark datasets, and they are not load-bearing for the fusion result. The weaknesses noted by the skeptic—absence of single-retriever baselines, CombSUM beating CombMNZ on several metric cells, promised t-tests not reported, and no repository link—are verifiability and consistency defects, not cases in which a claimed result equals its input by construction. The Section 1 complementarity premise is asserted rather than measured, but an unsupported assumption is not circular reasoning. No circular step can be exhibited with the required quote-and-reduction standard, so the score is 0.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The experimental design rests on a small number of hand-chosen hyperparameters (candidate pool of 128 per retriever, 80/20 Pareto split for head/tail labels) and on three unstated domain assumptions: that BM25 and BERT errors are complementary, that classical metasearch fusion algorithms transfer to two short XMTC candidate lists, and that the Pareto split is a meaningful partition for the reported tail/head metrics. No new entities are postulated.

free parameters (2)

Candidate pool size per retriever = 128 (64 tail + 64 head)
Section 4 sets the number of candidates to 64 tail and 64 head labels for both retrievers; no sensitivity analysis is given, and the final rankings depend on this choice.
Head/tail Pareto threshold = 80% tail / 20% head
Section 4 defines tail as the 80% least frequent labels and head as the remaining 20%; metrics are reported separately for these groups, so this arbitrary split shapes the headline results.

assumptions (5)

domain assumption The 80/20 Pareto split is a meaningful categorization of head versus tail labels for XMTC evaluation.
Section 4 introduces the threshold; all tail/head metrics depend on this arbitrary cutoff, which is not varied or validated.
domain assumption BM25 and BERT embedding retrievers produce complementary ranking signals for XMTC.
Section 1 asserts complementarity as the motivation for fusion, but the paper never measures ranking overlap or error correlation between the two retrievers.
domain assumption Classical metasearch fusion algorithms transfer unchanged to two short XMTC candidate lists.
The evaluated algorithms (Comb*, ISR, BordaFuse, Condorcet) were designed for aggregating many retrieval systems; their behavior with exactly two retrievers and 128 candidates is assumed, not analyzed.
standard math The two-sided paired Student's t-test is valid for the averaged nDCG and Precision values across the five folds.
Section 4 states the test is used, but no significance results are reported; the test's validity requires assumptions about fold independence and normality that are not discussed.
domain assumption nDCG@k and Precision@k with k in {1,5,10} are appropriate evaluation metrics for XMTC.
Section 4 justifies these k values by the average number of relevant labels, but no comparison to other metrics or k values is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Ranking-based Fusion Algorithms for Extreme Multi-label Text Classification (XMTC)." pith.science (2026). https://pith.science/paper/7FZQQ6HP

@misc{pith2026250703761,
  author       = {Pith},
  title        = {Pith review of: Ranking-based Fusion Algorithms for Extreme Multi-label Text Classification (XMTC)},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7FZQQ6HP}},
  note         = {Machine review of arXiv:2507.03761}
}

read the original abstract

In the context of Extreme Multi-label Text Classification (XMTC), where labels are assigned to text instances from a large label space, the long-tail distribution of labels presents a significant challenge. Labels can be broadly categorized into frequent, high-coverage \textbf{head labels} and infrequent, low-coverage \textbf{tail labels}, complicating the task of balancing effectiveness across all labels. To address this, combining predictions from multiple retrieval methods, such as sparse retrievers (e.g., BM25) and dense retrievers (e.g., fine-tuned BERT), offers a promising solution. The fusion of \textit{sparse} and \textit{dense} retrievers is motivated by the complementary ranking characteristics of these methods. Sparse retrievers compute relevance scores based on high-dimensional, bag-of-words representations, while dense retrievers utilize approximate nearest neighbor (ANN) algorithms on dense text and label embeddings within a shared embedding space. Rank-based fusion algorithms leverage these differences by combining the precise matching capabilities of sparse retrievers with the semantic richness of dense retrievers, thereby producing a final ranking that improves the effectiveness across both head and tail labels.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 18 canonical work pages

[1]

Arian Askari et al. 2023. Injecting the BM25 Score as Text Improves BERT-Based Re-rankers. In ECIR (Dublin, Ireland). Springer-Verlag, Berlin, Heidelberg, 66–83

work page 2023
[2]

Aslam and Mark H

Javed A. Aslam and Mark H. Montague. 2001. Models for Metasearch. In SIGIR 2001: Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, September 9-13, 2001, New Orleans, Louisiana, USA, W. Bruce Croft, David J. Harper, Donald H. Kraft, and Justin Zobel (Eds.). ACM, 275–284. https://doi.org/...

work page arXiv 2001
[3]

Elias Bassani. 2023. ranxhub: An Online Repository for Information Retrieval Runs. In SIGIR (Taipei,Taiwan). ACM, New York, NY, USA, 3210–3214

work page 2023
[4]

Elias Bassani and Luca Romelli. 2022. ranx.fuse: A Python Library for Metasearch. In CIKM. ACM, 4808–4812. https://doi.org/10.1145/3511808.3557207

arXiv 2022
[5]

Washington Cunha et al. 2023. An Effective, Efficient, and Scalable Confidence- Based Instance Selection Framework for Transformer-Based Text Classification. In SIGIR (Taipei, Taiwan). ACM, New York, NY, USA, 665–674

work page 2023
[6]

Washington Cunha, Vítor Mangaravite, Christian Gomes, Sérgio Canuto, Elaine Resende, Cecilia Nascimento, Felipe Viegas, Celso França, Wellington Santos Martins, Jussara M Almeida, et al. 2021. On the cost-effectiveness of neural and non-neural approaches and representations for text classification: A compre- hensive comparative study. Information Processi...

work page 2021
[7]

Washington Cunha, Alejandro Moreo, Andrea Esuli, Fabrizio Sebastiani, Leonardo Rocha, and Marcos André Gonçalves. 2024. A Noise-Oriented and Redundancy-Aware Instance Selection Framework. ACM Transactions on Infor- mation Systems (Nov. 2024)

work page 2024
[8]

Washington Cunha, Leonardo Rocha, and Marcos André Gonçalves. 2025. A thorough benchmark of automatic text classification: From traditional approaches to large language models. arXiv preprint arXiv:2504.01930 (2025)

work page Pith review arXiv 2025

Show all 21 references

[9]

Fox and Joseph A

Edward A. Fox and Joseph A. Shaw. 1993. Combination of Multiple Searches. In TREC (NIST Special Publication, Vol. 500-215) . National Institute of Standards and Technology (NIST), 243–252

1993
[10]

Vidit Jain, Jatin Prakash, Deepak Saini, Jian Jiao, Ramachandran Ramjee, and Manik Varma. 2023. Renee: END-TO-END TRAINING OF EXTREME CLASSI- FICATION MODELS. In Proceedings of Machine Learning and Systems , D. Song, M. Carbin, and T. Chen (Eds.), Vol. 5. Curan, 646–665

2023
[11]

Ting Jiang et al. 2021. Lightxml: Transformer with dynamic negative sampling for high-performance extreme multi-label text classification. In AAAI, Vol. 35. 7987–7994

2021
[12]

Jiajing Liu et al. 2023. A contrastive learning framework for safety information extraction in construction. Advanced Engineering Informatics 58 (2023), 102194

2023
[13]

Montague and Javed A

Mark H. Montague and Javed A. Aslam. 2002. Condorcet fusion for improved retrieval. In Proceedings of the 2002 ACM CIKM International Conference on Infor- mation and Knowledge Management, McLean, V A, USA, November 4-9, 2002. ACM, 538–548. https://doi.org/10.1145/584792.584881

2002
[14]

André Mourão, Flávio Martins, and João Magalhães. 2015. Multimodal medical information retrieval with unsupervised rank fusion. Comput. Medical Imaging Graph. 39 (2015), 35–45. https://doi.org/10.1016/j.compmedimag.2014.05.006

2015 doi
[15]

Andrea Pasin, Washington Cunha, Marcos André Gonçalves, and Nicola Ferro
[16]

Elena Renda and Umberto Straccia

M. Elena Renda and Umberto Straccia. 2003. Web metasearch: rank vs. score based rank aggregation methods. In Proceedings of the 2003 ACM Symposium on Conference acronym ’XX, June 03–05, 2018, Woodstock, NY França et al. Table 2: nDCG@k and Precision@k of Normalization Strategi...

2003
[17]

Wei Wei et al. 2022. Contrastive Meta Learning with Behavior Multiplicity for Recommendation. In WSDM. ACM, New York, NY, USA, 9 pages

2022
[18]

Hui Ye, Rajshekhar Sunderraman, and Shihao Ji. 2024. MatchXML: An Efficient Text-Label Matching Framework for Extreme Multi-Label Text Classification. IEEE Transactions on Knowledge and Data Engineering 36, 9 (2024), 4781–4793

2024
[19]

Ronghui You et al. 2019. AttentionXML: Label Tree-based Attention-Aware Deep Model for High-Performance Extreme Multi-Label Text Classification. InNeurIPS, Ranking-based Fusion Algorithms for Extreme Multi-label Text Classification (XMTC) Conference acronym ’XX, June 03–05, 20...

2019
[20]

Jiong Zhang et al . 2021. Fast Multi-Resolution Transformer Fine-tuning for Extreme Multi-label Text Classification. In NeurIPS, M. Ranzato et al . (Eds.), Vol. 34. Curran Associates, Inc., 7267–7280. Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009 Confe...

2021
[2024]

In Proceedings of the 2024 ACM SIGIR International Conference on Theory of Information Retrieval

A quantum annealing instance selection approach for efficient and effective transformer fine-tuning. In Proceedings of the 2024 ACM SIGIR International Conference on Theory of Information Retrieval . 205–214

2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Arian Askari et al. 2023. Injecting the BM25 Score as Text Improves BERT-Based Re-rankers. In ECIR (Dublin, Ireland). Springer-Verlag, Berlin, Heidelberg, 66–83

work page 2023

[2] [2]

Aslam and Mark H

Javed A. Aslam and Mark H. Montague. 2001. Models for Metasearch. In SIGIR 2001: Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, September 9-13, 2001, New Orleans, Louisiana, USA, W. Bruce Croft, David J. Harper, Donald H. Kraft, and Justin Zobel (Eds.). ACM, 275–284. https://doi.org/...

work page arXiv 2001

[3] [3]

Elias Bassani. 2023. ranxhub: An Online Repository for Information Retrieval Runs. In SIGIR (Taipei,Taiwan). ACM, New York, NY, USA, 3210–3214

work page 2023

[4] [4]

Elias Bassani and Luca Romelli. 2022. ranx.fuse: A Python Library for Metasearch. In CIKM. ACM, 4808–4812. https://doi.org/10.1145/3511808.3557207

arXiv 2022

[5] [5]

Washington Cunha et al. 2023. An Effective, Efficient, and Scalable Confidence- Based Instance Selection Framework for Transformer-Based Text Classification. In SIGIR (Taipei, Taiwan). ACM, New York, NY, USA, 665–674

work page 2023

[6] [6]

Washington Cunha, Vítor Mangaravite, Christian Gomes, Sérgio Canuto, Elaine Resende, Cecilia Nascimento, Felipe Viegas, Celso França, Wellington Santos Martins, Jussara M Almeida, et al. 2021. On the cost-effectiveness of neural and non-neural approaches and representations for text classification: A compre- hensive comparative study. Information Processi...

work page 2021

[7] [7]

Washington Cunha, Alejandro Moreo, Andrea Esuli, Fabrizio Sebastiani, Leonardo Rocha, and Marcos André Gonçalves. 2024. A Noise-Oriented and Redundancy-Aware Instance Selection Framework. ACM Transactions on Infor- mation Systems (Nov. 2024)

work page 2024

[8] [8]

Washington Cunha, Leonardo Rocha, and Marcos André Gonçalves. 2025. A thorough benchmark of automatic text classification: From traditional approaches to large language models. arXiv preprint arXiv:2504.01930 (2025)

work page Pith review arXiv 2025