Pith. sign in

REVIEW 6 major objections 6 minor 58 references

NLPnorth @ TalentCLEF 2025: Comparing Discriminative, Contrastive, and Prompt-Based Methods for Job Title and Skill Matching

T0 review · 6 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Zero-shot prompting with a 7-billion-parameter embedder wins multilingual job-title matching, while fine-tuned classification wins job-to-skill prediction.

desk verdict Useful shared-task comparison of three paradigms for job-title and skill matching, but the headline Task A result is a 0.013 MAP gap over 100 titles per language with no significance test—treat the ordering as provisional. read the letter →

arxiv 2506.19058 v1 pith:WLVORAVH submitted 2025-06-23 cs.CL

Mike Zhang , Rob van der Goot This is my paper · ORCID
classification cs.CL
keywords computationaljobmarketanalysisNLPforHumanResourcestitlematchingjob-skillclassificationcontrastivelearningpromptinglargelanguagemodels
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper reports a systematic comparison of three modeling paradigms on two tasks from the TalentCLEF 2025 shared task: multilingual job-title matching (Task A) and job-title-based skill prediction (Task B). The authors find that the best method differs by task: zero-shot prompting with a 7-billion-parameter instruction-tuned embedder (gte-Qwen2-7B-instruct) reaches the highest mean average precision of 0.492 on Task A test data averaged over English, Spanish, and German, while a fine-tuned classification model (mdeberta-v3-base) reaches 0.290 on Task B test data. The broader point is that there is no one-size-fits-all model for labor-market matching: prompt-based methods can rival supervised systems for title retrieval without task-specific training, but the direct supervised signal still wins for predicting skills. If correct, this gives practitioners a concrete task-by-task guide to choosing systems and suggests that instruction-tuned LLM embedders are now strong off-the-shelf tools for multilingual job title retrieval.

What carries the argument

The argument rests on three competing mechanisms. Classification reformulates ranking as binary relevance prediction: each query is paired with sampled negatives, an encoder is fine-tuned with cross-entropy through the MaChAmp toolkit, and the softmax score ranks candidates. Contrastive learning constructs positive pairs from ESCO (preferred title to description, preferred title to alternative title) and fine-tunes multilingual sentence embedders with the InfoNCE loss so that related titles and skills sit close in cosine space. Prompting skips training entirely: instruction-tuned LLM embedders (m-e5-large-instruct, Linq-Embed-Mistral, gte-Qwen2-7B-instruct) encode the query and candidates with a task prefix and rank by cosine similarity. The paper also builds a unified ESCO-derived corpus of titles, alternative labels, and multilingual descriptions, which supplies the training signal for the contrastive models.

What would settle it

Compute confidence intervals for the difference in MAP between gte-Qwen2-7B-instruct and m-e5-large contrastive on Task A by bootstrapping over the 100 gold-standard test titles per language (or by scoring the full 5,000-title test set if gold labels become available); if the interval crosses zero, the claim that prompting is the best title-matching method is falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a task-dependent ordering of methods. For multilingual job-title matching, the zero-shot prompt-based embedder gte-Qwen2-7B-instruct achieves the best test MAP of 0.492 averaged over English, Spanish, and German, outperforming the best fine-tuned contrastive model (m-e5-large, 0.480) and all classification models; it also leads on English–Spanish and English–Chinese cross-lingual transfer. For job-title-based skill prediction, fine-tuned classification with mdeberta-v3-base achieves the best test MAP of 0.290, ahead of contrastive and prompting alternatives. The paper concludes that instruction-tuned prompting is effective for title matching and that classification excels at skill prediction, so model choice should be task-specific rather than universal.

Load-bearing premise

The paper's headline claim that zero-shot prompting is the best Task A method depends on the assumption that the 100 gold-standard test titles per language are representative and that the reported margin over fine-tuned contrastive learning (0.493 vs 0.480 MAP) is real rather than sampling noise, since no variance estimates or significance tests are reported.

Editorial extensions

If this is right

  • For multilingual job-title matching, zero-shot prompting with a 7B instruction-tuned embedder is competitive with, and here superior to, fine-tuned contrastive systems, so teams without training data or compute for fine-tuning can still deploy a strong title matcher.
  • For skill prediction, the direct supervised signal from classification matters more: fine-tuned classification should be the default until prompting methods close the gap.
  • Cross-lingual transfer works best through the largest prompt-based embedder, which leads on English–Spanish and English–Chinese pairs even though Chinese was unseen during training.
  • Model choice should be task-specific: the winning approach for title matching is not the winning approach for skill prediction, so evaluations should report the two tasks separately.
  • Skill prediction is harder than title matching across all methods, as evidenced by the much lower MAP scores on Task B.

Reading between the lines

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

  • The one-point test gap between prompting (0.493) and contrastive (0.480) on Task A is not tested for significance; with only 100 gold titles per language, a bootstrap or paired test could plausibly erase the gap, so the headline ordering should be treated as provisional.
  • Read the abstract's 'largest models win both tasks' as applying to Task A: in Task B the best classifier, mdeberta-v3-base, is a 276M-parameter model, so model size is not the consistent driver across both tasks.
  • The strong zero-shot cross-lingual results suggest a cheap extension: prompting with instruction-tuned embedders could be tested on languages absent from ESCO altogether, not just Chinese, to see whether the transfer advantage holds.
  • Because the contrastive models were the only ones trained on ESCO descriptions, the comparison between contrastive and prompting partly confounds training signal with method; a prompt-based model given the same descriptions might close the gap further.
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.

Referee Report

6 major / 6 minor

Summary. This working note describes NLPnorth's submissions to TalentCLEF 2025 Task A (multilingual job-title matching) and Task B (job-title-based skill prediction). The authors compare fine-tuned classification (MaChAmp with cross-entropy), fine-tuned contrastive learning (InfoNCE on ESCO-derived title/description pairs), and zero-shot prompting with instruction-tuned embedders, the largest being gte-Qwen2-7B-instruct. On Task A test data they report 0.492 MAP averaged over English, Spanish, and German for zero-shot prompting, marginally ahead of 0.480 for fine-tuned contrastive matching; on Task B they report 0.290 MAP for fine-tuned classification with mdeberta-v3-base versus 0.283 for prompting. The paper concludes that model choice should be task-specific rather than one-size-fits-all, and that prompting is competitive for title matching while discriminative training leads for skill prediction.

Significance. Taken at face value, the paper offers a useful side-by-side comparison of three dominant NLP paradigms on an external shared-task benchmark, with the notable finding that a 7B zero-shot embedder is competitive with fine-tuned specialist models on multilingual title retrieval. The authors release code and preprocessing scripts and construct a unified ESCO-derived training corpus, which supports reproducibility; hyperparameters are selected on development data rather than on test labels. The main contribution is as a system description, and the empirical rankings are directly supported by the reported MAP scores. However, the headline differences are small, are estimated on a 100-query-per-language test subset, and are presented without any uncertainty quantification, so the paper's central conclusions should be treated as provisional pending additional statistical analysis.

major comments (6)
  1. [Section 4.1, Tables 2 and 3] The headline Task A result—zero-shot gte-Qwen2-7B-instruct at 0.492/0.493 MAP versus fine-tuned contrastive m-e5-large at 0.480—is estimated on a gold subset of only 100 titles per language (Section 2.1) and is reported without confidence intervals or significance tests. The per-language breakdown in Table 3 shows the prompting model trailing on German (0.442 vs 0.446), so the 0.013 average gap could plausibly be sampling noise. Please add paired bootstrap or per-query AP significance testing over the roughly 300 queries, and revise the abstract and conclusions to match the outcome. Please also reconcile the reporting discrepancy: the abstract gives 0.492 for the en/es/de average, while Table 3 gives 0.493 for the same average, which does not match the arithmetic of the row's per-language values.
  2. [Footnote 2, Section 4.1] The best Task A prompt-based run (gte-Qwen2-7B-instruct) was obtained after the submission deadline; the officially submitted model was m-e5-large-instruct. Because the abstract and conclusion present the post-deadline run as the paper's best result without mentioning this, the paper conflates the official shared-task ranking with exploratory post-deadline experimentation. Please state this caveat in the abstract and clearly label post-deadline rows in Tables 2 and 3.
  3. [Abstract, Section 1, Section 4.2, Table 4] The abstract claims that 'the largest multilingual language models perform best for both tasks,' but Section 4.2 states that model size has 'a less clear impact' and that 'the smallest language model performs best' for Task B, and Table 4 shows the best test model is mdeberta-v3-base with 276M parameters. Section 1 also states that 'contrastive learning excels on multilingual title matching,' whereas Section 4.1 and the abstract crown prompting. These internal inconsistencies need to be aligned with the actual results before the paper's task-specific conclusion can be evaluated.
  4. [Section 4.1 and Section 4.2] The text says 'For the test predictions, we select the best performing model from each category.' Please clarify explicitly whether this selection was made on development MAP or on test MAP. If test labels influenced model selection, the reported test scores are circular; if development was used, state this in the text. The same clarification is needed for both tasks.
  5. [Section 4.2, Table 4] The Task B conclusion that classification 'excels' over prompting rests on a test MAP gap of only 0.007 (0.290 vs 0.283). Although the validation gap is larger (0.267 vs 0.222), no significance test accompanies the test comparison, so the stated superiority is not established. Please add a significance or stability analysis or soften the claim to 'competitive.'
  6. [Section 4.1, Table 2] The test-set model selection wording and the interleaved validation/test columns in Table 2 make it hard to determine which models were actually evaluated on the test split and which numbers are validation-only. In particular, the gte-Qwen2-7B-instruct test scores are post-deadline, and it is unclear whether the 'best performing model from each category' was chosen using development or test labels. Please make the selection protocol and the table structure explicit.
minor comments (6)
  1. [Section 4.1, Section 4.2] Section 4.1 contains the typo 'bath datasplits' (should be 'both data splits'), and Section 4.2 contains 'a the smallest language model' (should be 'the smallest language model').
  2. [Tables 2 and 4] The tables interleave validation and test columns without clear group headers, making it difficult to tell which models were evaluated on the test split. Add explicit column groups and use a dash for unmeasured cells.
  3. [Section 5, Tables 5 and 6] The per-category analysis covers only the mapped subset, since Table 5 shows that 77.4--87.8% of validation titles are unmapped to ESCO major groups. The text's statement that 'most models do not perform well' on unmapped titles is hard to reconcile with the 'Unmapped Titles' row in Table 6, which shows MAP values of 0.426--0.617. Please clarify what that row represents and discuss the coverage limitation.
  4. [Section 3.3] The instruction-tuned embedders gte-Qwen2-7B-instruct and Linq-Embed-Mistral are not cited; please add references for both models.
  5. [Section 3.1] For Task B, the statement that 'the optimal negative ratio for taskB was 1:1' does not say whether this ratio was selected on development data; please state the selection protocol.
  6. [Table 3] The 'Avg. (en,es,de)' column value of 0.493 for gte-Qwen2-7B-instruct is inconsistent with the abstract's 0.492 and with the arithmetic of the row's per-language values (0.537, 0.496, 0.442 average to 0.492). Correct the rounding or the table entry.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: all reported results are measured MAP scores on the external TalentCLEF test data, with no derivation step that reduces to its own inputs.

full rationale

The paper's central claims are empirical rankings of systems on shared-task test sets, not quantities derived from the paper's own definitions. Task A and Task B results (Tables 2 and 4) are reported mean average precision values computed against gold-standard test annotations provided by TalentCLEF, which is an external benchmark. The only self-citations are escoxlm-r [53] as one of the compared models and MaChAmp [52] as the training toolkit; neither is used as load-bearing evidence for the performance ordering. Hyperparameters such as negative ratio, batch size, and learning rate are tuned on development data before test evaluation, and the test MAP values are then reported for the selected models per category, so no fitted parameter is renamed as a prediction. Footnote 2 states that the best Task A prompting results were obtained after the submission deadline and that the m-e5-large-instruct model was the officially submitted one; this is a timing and practical-claim limitation, not a circularity. The skeptic's concern about the absence of significance tests and the small 100-title test subset is a statistical robustness issue, not evidence that any result reduces by construction to its inputs. No self-definitional relation, no imported uniqueness theorem, no ansatz smuggled in via citation, and no renaming of a known result as an organization principle appear in the text. Because the evaluation is anchored to an external shared-task benchmark with provided gold labels, the derivation chain is self-contained and the circularity score is 0.

Assumptions & free parameters 7 free parameters · 3 assumptions · 0 invented entities

The paper contributes benchmark evidence rather than a derivation, so the ledger contains training hyperparameters tuned on development data and domain assumptions about the ESCO taxonomy and the shared-task annotations. No new theoretical entities are introduced; the 'unified ESCO corpus' is a dataset construction, not a new postulated entity.

free parameters (7)
  • Task A classification negative sampling ratio = 1:2
    Selected on English development data among ratios 1:1, 1:2, and 1:5 (Section 3.1).
  • Task B classification negative sampling ratio = 1:1
    Reported as the optimal negative ratio for Task B (Section 3.1).
  • Contrastive number of in-batch negatives k = 16
    Optimal value from the sweep k in {1,2,5,10,15,16,20,32} (Section 3.2).
  • Contrastive batch size = 32
    Optimal value from the sweep {16,32,64} (Section 3.2).
  • Contrastive learning rate = 2e-6
    Optimal value from the sweep {1e-4, 5e-5, 2e-5, 2e-6} (Section 3.2).
  • Classification number of epochs = 3
    Reduced from the default 20 after observing early convergence (Section 3.1).
  • Prompt prefixes for Task A and Task B = Task A: 'Given a job title, find the most relevant job titles.' Task B: 'Given a job title, find the most relevant…
    Hand-crafted task descriptions with no systematic ablation (Section 3.3).
assumptions (3)
  • domain assumption ESCO preferred titles, alternative titles, and multilingual descriptions form reliable positive pairs for job and skill relatedness.
    Used to build all contrastive training pairs in Section 3.2; if these pairs are noisy, the contrastive results in Tables 2 and 4 are not trustworthy.
  • domain assumption The TalentCLEF gold annotations and the 100-title test subset per language are correct and representative.
    All MAP scores in Section 4 are computed against these labels; a small or noisy test subset could change which method appears best.
  • domain assumption Standard transformer fine-tuning and sentence-embedding behavior transfers to this domain as reported by the original model papers.
    The paper does not re-derive or verify model behavior; it relies on pretrained multilingual encoders and InfoNCE training as background tools.

how reviews work

0 comments
Cite this review

Pith. "Pith review of NLPnorth @ TalentCLEF 2025: Comparing Discriminative, Contrastive, and Prompt-Based Methods for Job Title and Skill Matching." pith.science (2026). https://pith.science/paper/WLVORAVH

@misc{pith2026250619058,
  author       = {Pith},
  title        = {Pith review of: NLPnorth @ TalentCLEF 2025: Comparing Discriminative, Contrastive, and Prompt-Based Methods for Job Title and Skill Matching},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WLVORAVH}},
  note         = {Machine review of arXiv:2506.19058}
}
abstract

Matching job titles is a highly relevant task in the computational job market domain, as it improves e.g., automatic candidate matching, career path prediction, and job market analysis. Furthermore, aligning job titles to job skills can be considered an extension to this task, with similar relevance for the same downstream tasks. In this report, we outline NLPnorth's submission to TalentCLEF 2025, which includes both of these tasks: Multilingual Job Title Matching, and Job Title-Based Skill Prediction. For both tasks we compare (fine-tuned) classification-based, (fine-tuned) contrastive-based, and prompting methods. We observe that for Task A, our prompting approach performs best with an average of 0.492 mean average precision (MAP) on test data, averaged over English, Spanish, and German. For Task B, we obtain an MAP of 0.290 on test data with our fine-tuned classification-based approach. Additionally, we made use of extra data by pulling all the language-specific titles and corresponding \emph{descriptions} from ESCO for each job and skill. Overall, we find that the largest multilingual language models perform best for both tasks. Per the provisional results and only counting the unique teams, the ranking on Task A is 5$^{\text{th}}$/20 and for Task B 3$^{\text{rd}}$/14.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 43 canonical work pages

  1. [1]

    H. Lasi, P. Fettke, H.-G. Kemper, T. Feld, M. Hoffmann, Industry 4.0, Business & information systems engineering 6 (2014) 239–242

    2014

  2. [2]

    Schwab, The fourth industrial revolution, Currency, 2017

    K. Schwab, The fourth industrial revolution, Currency, 2017

    work page 2017

  3. [3]

    Accessed: 2023-10-27

    European Commission, Industry 5.0: Towards more sustainable, resilient and human-centric industry, https://research-and-innovation.ec.europa.eu/news/all-research-and-innovation-news/ industry-50-towards-more-sustainable-resilient-and-human-centric-industry-2021-01-07_en, 2021. Accessed: 2023-10-27

    work page 2021

  4. [4]

    Eloundou, S

    T. Eloundou, S. Manning, P. Mishkin, D. Rock, Gpts are gpts: An early look at the labor market impact potential of large language models, ArXiv preprint abs/2303.10130 (2023). URL: https://arxiv.org/abs/ 2303.10130

    arXiv 2023

  5. [5]

    D. H. Autor, D. Dorn, The growth of low-skill service jobs and the polarization of the us labor market, American economic review 103 (2013) 1553–1597. URL: https://www.aeaweb.org/articles?id=10.1257/ aer.103.5.1553

    work page 2013

  6. [6]

    D. H. Autor, F. Levy, R. J. Murnane, The skill content of recent technological change: An empirical exploration, The Quarterly journal of economics 118 (2003) 1279–1333. URL: https://academic.oup. com/qje/article-abstract/118/4/1279/1925105?login=false

    work page 2003

  7. [7]

    Brynjolfsson, A

    E. Brynjolfsson, A. McAfee, Race against the machine: How the digital revolution is accelerating inno- vation, driving productivity, and irreversibly transforming employment and the economy, Brynjolfsson and McAfee, 2011

    work page 2011

  8. [8]

    Brynjolfsson, A

    E. Brynjolfsson, A. McAfee, The second machine age: Work, progress, and prosperity in a time of brilliant technologies, WW Norton & Company, 2014

    2014

Show all 58 references
  1. [9]

    Balog, Y

    K. Balog, Y. Fang, M. De Rijke, P. Serdyukov, L. Si, et al., Expertise retrieval, Foundations and Trends® in Information Retrieval 6 (2012) 127–256

    2012

  2. [10]

    Otani, N

    N. Otani, N. Bhutani, E. Hruschka, Natural language processing for human resources: A survey, ArXiv preprint abs/2410.16498 (2024). URL: https://arxiv.org/abs/2410.16498

    2024 arXiv

  3. [11]

    Senger, M

    E. Senger, M. Zhang, R. van der Goot, B. Plank, Deep learning-based computational job market analysis: A survey on skill extraction and classification from job postings, in: E. Hruschka, T. Lake, N. Otani, T. Mitchell (Eds.), Proceedings of the First Workshop on Natural Langua...

    2024

  4. [12]

    Zhang, Computational Job Market Analysis: with Natural Language Processing, 2024

    M. Zhang, Computational Job Market Analysis: with Natural Language Processing, 2024

    2024

  5. [13]

    Sayfullina, E

    L. Sayfullina, E. Malmi, J. Kannala, Learning representations for soft skill matching, in: International Conference on Analysis of Images, Social Networks and Texts, 2018, pp. 141–152

    2018

  6. [14]

    Bhola, K

    A. Bhola, K. Halder, A. Prasad, M.-Y. Kan, Retrieving skills from job descriptions: A language model based extreme multi-label classification framework, in: D. Scott, N. Bel, C. Zong (Eds.), Proceedings of the 28th International Conference on Computational Linguistics, Interna...

    2020 doi

  7. [15]

    Khaouja, G

    I. Khaouja, G. Mezzour, I. Kassou, Unsupervised skill identification from job ads, in: 2021 IEEE 22nd International Conference on Information Reuse and Integration for Data Science (IRI), IEEE, 2021, pp. 147–151

    2021

  8. [16]

    Zhang, K

    M. Zhang, K. Jensen, S. Sonniks, B. Plank, SkillSpan: Hard and soft skill extraction from English job postings, in: M. Carpuat, M.-C. de Marneffe, I. V. Meza Ruiz (Eds.), Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Ling...

    2022 doi

  9. [17]

    Zhang, K

    M. Zhang, K. N. Jensen, B. Plank, Kompetencer: Fine-grained skill classification in danish job postings via distant supervision and transfer learning, in: Proceedings of the Language Resources and Evaluation Conference, European Language Resources Association, Marseille, Franc...

    2022

  10. [18]

    Zhang, K

    M. Zhang, K. N. Jensen, R. van der Goot, B. Plank, Skill extraction from job postings using weak supervision, in: Proceedings of RecSysHR’22, RecSysHR’22, 2022

    2022

  11. [19]

    Green, D

    T. Green, D. Maynard, C. Lin, Development of a benchmark corpus to support entity recognition in job descriptions, in: Proceedings of the Language Resources and Evaluation Conference, European Language Resources Association, Marseille, France, 2022, pp. 1201–1208. URL: https:/...

    2022

  12. [20]

    Gnehm, E

    A.-S. Gnehm, E. Bühlmann, S. Clematide, Evaluation of transfer learning and domain adaptation for analyzing german-speaking job advertisements, in: Proceedings of the Language Resources and Evaluation Conference, European Language Resources Association, Marseille, France, 2022...

    2022

  13. [21]

    Nguyen, M

    K. Nguyen, M. Zhang, S. Montariol, A. Bosselut, Rethinking skill extraction in the job market domain using large language models, in: Proceedings of the First Workshop on Natural Language Processing for Human Resources (NLP4HR 2024), Association for Computational Linguistics, ...

    2024

  14. [22]

    Herandi, Y

    A. Herandi, Y. Li, Z. Liu, X. Hu, X. Cai, Skill-llm: Repurposing general-purpose llms for skill extraction, ArXiv preprint abs/2410.12052 (2024). URL: https://arxiv.org/abs/2410.12052

    2024 arXiv

  15. [23]

    Vásquez-Rodríguez, S

    L. Vásquez-Rodríguez, S. M. Bertrand Audrin, S. Galli, J. Rogenhofer, J. N. Cusa, L. van der Plas, Hardware-effective approaches for skill extraction in job offers and resumes (2024)

    2024

  16. [24]

    Zhang, R

    M. Zhang, R. v. d. Goot, M.-Y. Kan, B. Plank, NNOSE: Nearest neighbor occupational skill extraction, in: Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Linguisti...

    2024

  17. [25]

    Kavas, M

    H. Kavas, M. Serra-Vidal, L. Wanner, Multilingual skill extraction for job vacancy–job seeker matching in knowledge graphs, in: Proceedings of the Workshop on Generative AI and Knowledge Graphs (GenAIK), 2025, pp. 146–155

    2025

  18. [26]

    le Vrang, A

    M. le Vrang, A. Papantoniou, E. Pauwels, P. Fannes, D. Vandensteen, J. De Smedt, Esco: Boosting job matching in europe with semantic interoperability, Computer 47 (2014) 57–64

    2014

  19. [27]

    Giabelli, L

    A. Giabelli, L. Malandri, F. Mercorio, M. Mezzanzanica, Graphlmi: A data driven system for exploring labor market information through graph databases, Multimedia Tools and Applications (2020) 1–30

    2020

  20. [28]

    de Groot, J

    M. de Groot, J. Schutte, D. Graus, Job posting-enriched knowledge graph for skills-based matching, 2021.arXiv:2109.02554

    2021 arXiv

  21. [29]

    Yazdanian, R

    R. Yazdanian, R. L. Davis, X. Guo, F. Lim, P. Dillenbourg, M.-Y. Kan, On the radar: Predicting near-future surges in skills’ hiring demand to provide early warning to educators, Computers and Education: Artificial Intelligence (2021) 100043

    2021

  22. [30]

    Decorte, J

    J.-J. Decorte, J. Van Hautte, J. Deleu, C. Develder, Demeester, Design of negative sampling strategies for distantly supervised skill extraction, ArXiv preprint abs/2209.05987 (2022). URL: https://arxiv.org/ abs/2209.05987

    2022 arXiv

  23. [31]

    Clavié, G

    B. Clavié, G. Soulié, Large language models as batteries-included zero-shot esco skills matchers, ArXiv preprint abs/2307.03539 (2023). URL: https://arxiv.org/abs/2307.03539

    2023 arXiv

  24. [32]

    Decorte, J

    J.-J. Decorte, J. V. Hautte, T. Demeester, C. Develder, Skillmatch: Evaluating self-supervised learning of skill relatedness, ArXiv preprint abs/2410.05006 (2024). URL: https://arxiv.org/abs/2410.05006

    2024 arXiv

  25. [33]

    Magron, A

    A. Magron, A. Dai, M. Zhang, S. Montariol, A. Bosselut, JobSkape: A framework for generating synthetic job postings to enhance skill matching, in: E. Hruschka, T. Lake, N. Otani, T. Mitchell (Eds.), Proceedings of the First Workshop on Natural Language Processing for Human Res...

    2024

  26. [34]

    Gavrilescu, F

    M. Gavrilescu, F. Leon, A.-A. Minea, Techniques for transversal skill classification and relevant keyword extraction from job advertisements, Information 16 (2025) 167

    2025

  27. [35]

    Elias, M

    P. Elias, M. Birch, et al., Soc2010: revision of the standard occupational classification, Economic & Labour Market Review 4 (2010) 48–55

    2010

  28. [36]

    Elias, Occupational Classification (ISCO-88): Concepts, Methods, Reliability, Validity and Cross- National Comparability, Technical Report, OECD Publishing, 1997

    P. Elias, Occupational Classification (ISCO-88): Concepts, Methods, Reliability, Validity and Cross- National Comparability, Technical Report, OECD Publishing, 1997

    1997

  29. [37]

    Decorte, J

    J.-J. Decorte, J. Van Hautte, T. Demeester, C. Develder, Jobbert: Understanding job titles through skills, ArXiv preprint abs/2109.09605 (2021). URL: https://arxiv.org/abs/2109.09605

    2021 arXiv

  30. [38]

    Javed, Q

    F. Javed, Q. Luo, M. McNair, F. Jacob, M. Zhao, T. S. Kang, Carotene: A job title classification system for the online recruitment domain, in: 2015 IEEE First International Conference on Big Data Computing Service and Applications, IEEE, 2015, pp. 286–293

    2015

  31. [39]

    Javed, M

    F. Javed, M. McNair, F. Jacob, M. Zhao, Towards a job title classification system, ArXiv preprint abs/1606.00917 (2016). URL: https://arxiv.org/abs/1606.00917

    2016 arXiv

  32. [40]

    Javed, P

    F. Javed, P. Hoang, T. Mahoney, M. McNair, Large-scale occupational skills normalization for online recruitment, in: S. P. Singh, S. Markovitch (Eds.), Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence, February 4-9, 2017, San Francisco, California, US...

    2017

  33. [41]

    Retyk, L

    F. Retyk, L. Gasco, C. P. Carrino, D. Deniz, R. Zbib, Melo: An evaluation benchmark for multilingual entity linking of occupations, ArXiv preprint abs/2410.08319 (2024). URL: https://arxiv.org/abs/2410. 08319

    2024 arXiv

  34. [42]

    X. Liu, Y. Wang, Q. Dong, X. Lu, Job title prediction as a dual task of expertise prediction in open source software, in: Joint European Conference on Machine Learning and Knowledge Discovery in Databases, Springer, 2024, pp. 381–396

    2024

  35. [43]

    Malherbe, M

    E. Malherbe, M. Aufaure, Bridge the terminology gap between recruiters and candidates: A multilingual skills base built from social media and linked data, in: R. Kumar, J. Caverlee, H. Tong (Eds.), 2016 IEEE/ACM International Conference on Advances in Social Networks Analysis ...

    2016

  36. [44]

    E. M. Sibarani, S. Scerri, C. Morales, S. Auer, D. Collarana, Ontology-guided job market demand analysis: a cross-sectional study for the data science field, in: Proceedings of the 13th International Conference on Semantic Systems, 2017, pp. 25–32

    2017

  37. [45]

    Decorte, J

    J.-J. Decorte, J. Van Hautte, J. Deleu, C. Develder, T. Demeester, Career path prediction using resume representation learning and skill-based matching, in: RecSys in HR2023: the 3rd Workshop on Recommender Systems for Human Resources, in conjunction with the 17th ACM Conferen...

    2023

  38. [46]

    Senger, Y

    E. Senger, Y. Campbell, R. van der Goot, B. Plank, KARRIEREWEGE: A large scale career path prediction dataset, in: O. Rambow, L. Wanner, M. Apidianaki, H. Al-Khalifa, B. D. Eugenio, S. Schockaert, K. Darwish, A. Agarwal (Eds.), Proceedings of the 31st International Conference ...

    2025

  39. [47]

    URL: https://esco.ec.europa.eu/en/about-esco/data-science-and-esco/ machine-learning-assisted-mapping-multilingual-occupational-data-esco-part-1

    ESCO, Machine Learning Assisted Mapping of Multilingual Occupational Data to ESCO (Part 1), 2022. URL: https://esco.ec.europa.eu/en/about-esco/data-science-and-esco/ machine-learning-assisted-mapping-multilingual-occupational-data-esco-part-1

    2022

  40. [48]

    Vrolijk, D

    J. Vrolijk, D. Graus, Enhancing plm performance on labour market tasks via instruction-based finetuning and prompt-tuning with rules, ArXiv preprint abs/2308.16770 (2023). URL: https://arxiv. org/abs/2308.16770

    2023 arXiv

  41. [49]

    Zhang, R

    M. Zhang, R. v. d. Goot, B. Plank, Entity linking in the job market domain, in: Findings of the Association for Computational Linguistics: EACL 2024, Association for Computational Linguistics, St. Julian’s, Malta, 2024, pp. 410–419. URL: https://aclanthology.org/2024.findings-eacl.28

    2024

  42. [50]

    Rosenberger, L

    J. Rosenberger, L. Wolfrum, S. Weinzierl, M. Kraus, P. Zschech, Careerbert: Matching resumes to esco jobs in a shared embedding space for generic job recommendations, Expert Systems with Applications 275 (2025) 127043

    2025

  43. [51]

    Gasco, H

    L. Gasco, H. Fabregat, L. García-Sardiña, P. Estrella, D. Deniz, A. Rodrigo, R. Zbib, Overview of the TalentCLEF 2025 Shared Task: Skill and Job Title Intelligence for Human Capital Management, in: International Conference of the Cross-Language Evaluation Forum for European La...

    2025

  44. [52]

    van der Goot, A

    R. van der Goot, A. Üstün, A. Ramponi, I. Sharaf, B. Plank, Massive choice, ample tasks (MaChAmp): A toolkit for multi-task learning in NLP, in: D. Gkatzia, D. Seddah (Eds.), Proceedings of the 16th Conference of the European Chapter of the Association for Computational Lingui...

    2021 doi

  45. [53]

    Zhang, R

    M. Zhang, R. van der Goot, B. Plank, ESCOXLM-R: Multilingual Taxonomy-driven Pre-training for the Job Market Domain, in: Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), Association for Computational Linguistics, ...

    2023 doi

  46. [54]

    Zhang, Y

    Z. Zhang, Y. Gao, J.-G. Lou, 𝑒5: Zero-shot hierarchical table analysis using augmented LLMs via explain, extract, execute, exhibit and extrapolate, in: K. Duh, H. Gomez, S. Bethard (Eds.), Proceedings of the 2024 Conference of the North American Chapter of the Association for ...

    2024

  47. [55]

    P. He, J. Gao, W. Chen, Debertav3: Improving deberta using electra-style pre-training with gradient- disentangled embedding sharing, 2021.arXiv:2111.09543

    2021 arXiv

  48. [56]

    Reimers, I

    N. Reimers, I. Gurevych, Making monolingual sentence embeddings multilingual using knowledge distillation, in: Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), Association for Computational Linguistics, Online, 2020, pp. 4512–4525...

    2020 doi

  49. [57]

    Howard, S

    J. Howard, S. Ruder, Universal language model fine-tuning for text classification, in: I. Gurevych, Y. Miyao (Eds.), Proceedings of the 56th Annual Meeting of the Association for Computational Lin- guistics (Volume 1: Long Papers), Association for Computational Linguistics, Me...

    2018 doi

  50. [58]

    A. v. d. Oord, Y. Li, O. Vinyals, Representation learning with contrastive predictive coding, ArXiv preprint abs/1807.03748 (2018). URL: https://arxiv.org/abs/1807.03748

    2018 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.