Elamur Research⌕
☰
home / papers / 2026 / Corpus and Baselines for Distinguishing Authentic, AI-Generated, and AI-Enhanced Resumes

Corpus and Baselines for Distinguishing Authentic, AI-Generated, and AI-Enhanced Resumes

Summary

Job seekers can now use generative AI to write a resume straight from a job posting or to polish an existing one. Such documents often score highly in keyword-driven applicant tracking systems, which puts genuine candidates at a disadvantage. General AI-text detectors are built for continuous prose, so they struggle with the fragmented, list-heavy layout of resumes.

The authors present what they describe as the first annotated corpus for three-way resume classification. It covers five IT-sector job descriptions, with equal numbers of three resume types:

  • Authentic: real pre-2019 submissions supplied by the recruiting firm TECKpert, semi-automatically stripped of PII.
  • AI-generated: produced by prompting LLMs with only the job posting.
  • AI-enhanced: LLM rewrites of the authentic resumes.

As baselines, they train classic classifiers on TF-IDF and stylometric features and fine-tune BERT and Longformer. XGBoost performs best, ahead of the neural models. Two commercial detectors perform poorly on the same data. The released corpus gives researchers a public benchmark, and the results caution against using uncalibrated detectors in hiring.

Key points

  • Corpus: 420 resumes (140 per class) tied to five IT job descriptions, publicly released (DOI 10.34703/gzx1-9v95/HTRD9P).
  • Features: TF-IDF word and character n-grams combined with style cues. These cover length, punctuation, casing, contractions and TTR, along with sentiment, readability indices, hapax legomena and how unique the n-grams are.
  • Best model: XGBoost reaches 95.29% accuracy and a macro F1 of 0.953, outperforming Logistic Regression, SVM, Random Forest, BERT and Longformer.
  • Class differences:
    • AI-generated resumes use shorter sentences of more even length and rarely use contractions.
    • AI-enhanced resumes show the top TTR and uniqueness scores.
    • Authentic resumes vary the most on every feature.
  • Off-the-shelf detectors: Originality gets 55.7% overall (71/140 authentic, 81/140 generated, 82/140 enhanced). Writer manages only 25.0% and fails most on AI-enhanced resumes, which points to domain shift.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/4o69wu63q5hq

@inproceedings{loizidou2026corpus,
  title = {{Corpus and Baselines for Distinguishing Authentic, AI-Generated, and AI-Enhanced Resumes}},
  author = {Loizidou, Andrea and Sharma, Anshu Kiran and Esquivel, Adrian and Finlayson, Mark A. and Ocal, Mustafa},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/4o69wu63q5hq},
  url = {https://doi.org/10.63317/4o69wu63q5hq}
}

References

47 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sahar Abdelnabi and Mario Fritz. 2021. Adversarial watermarking transformer: Towards tracing text provenance with data hiding. In 2021 IEEE Symposium on Security and Privacy (SP), pages 121–140. IEEE. find
  2. Arslan Akram. 2023. An empirical study of ai generated text detection tools. arXiv preprint arXiv:2310.01423. — [arXiv:2310.01423](https://arxiv.org/abs/2310.01423) find
  3. Lena Armstrong and Danaé Metaxa. 2025. Navigating automated hiring: Perceptions, strategy use, and outcomes among young job seekers. Proceedings of the ACM on Human-Computer Interaction, 9(2):1–26. find
  4. Prafulla Bafna, Shailaja Shirwaikar, and Dhanya Pramod. 2019. Task recommender system using semantic clustering to identify the right personnel. VINE Journal of Information and Knowledge Management Systems, 49(2):181–199. find
  5. Anton Bakhtin, Sam Gross, Myle Ott, Yuntian Deng, Marc’Aurelio Ranzato, and Arthur Szlam. 2019. Real or fake? learning to discriminate machine from human generated text. arXiv preprint arXiv:1906.03351. — [arXiv:1906.03351](https://arxiv.org/abs/1906.03351) find
  6. Barbara K Brown and Michael A Campion. 1994. Biodata phenomenology: Recruiters’ perceptions and use of biographical information in resume screening. Journal of Applied Psychology, 79(6):897. find
  7. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901. find
  8. Elizabeth Clark, Tal August, Sofia Serrano, Nikita Haduong, Suchin Gururangan, and Noah A Smith. 2021. All that’s’ human’is not gold: Evaluating human evaluation of generated text. arXiv preprint arXiv:2107.00061. find
  9. Evan Crothers, Nathalie Japkowicz, Herna Viktor, and Paula Branco. 2022. Adversarial robustness of neural-statistical features in detection of generative transformers. In 2022 international joint conference on neural networks (IJCNN), pages 1–8. IEEE. find
  10. Asmita Deshmukh and Anjali Raut. 2025. Applying bert-based nlp for automated resume screening and candidate ranking. Annals of Data Science, 12(2):591–603. find
  11. Evanthia Faliagka, Athanasios Tsakalidis, and Giannis Tzimas. 2012. An integrated e-recruitment system for automated personality mining and applicant ranking. Internet research, 22(5):551– 568. find
  12. Rudolph Flesch. 1948. A new readability yardstick. Journal of applied psychology, 32(3):221. find
  13. Leon Fröhling and Arkaitz Zubiaga. 2021. Feature-based detection of automated language models: tackling gpt-2, gpt-3 and grover. PeerJ Computer Science, 7:e443. find
  14. Joseph B Fuller and Manjari Raman. 2022. Building from the bottom up. Harvard Business School, January. find
  15. Chenxi Gu, Chengsong Huang, Xiaoqing Zheng, Kai-Wei Chang, and Cho-Jui Hsieh. 2022. Watermarking pre-trained language models with backdooring. arXiv preprint arXiv:2210.07543. — [arXiv:2210.07543](https://arxiv.org/abs/2210.07543) find
  16. Abhimanyu Hans, Avi Schwarzschild, Valeriia Cherepanova, Hamid Kazemi, Aniruddha Saha, Micah Goldblum, Jonas Geiping, and Tom Goldstein. 2024. Spotting llms with binoculars: Zero-shot detection of machine-generated text. arXiv preprint arXiv:2401.12070. find
  17. Abe Bohan Hou, Jingyu Zhang, Tianxing He, Yichen Wang, Yung-Sung Chuang, Hongwei Wang, Lingfeng Shen, Benjamin Van Durme, Daniel Khashabi, and Yulia Tsvetkov. 2023. Semstamp: A semantic watermark with paraphrastic robustness for text generation. arXiv preprint arXiv:2310.03991. find
  18. Xiaomeng Hu, Pin-Yu Chen, and Tsung-Yi Ho. 2023. Radar: Robust ai-text detection via adversarial learning. Advances in neural information processing systems, 36:15077–15095. find
  19. Clayton Hutto and Eric Gilbert. 2014. Vader: A parsimonious rule-based model for sentiment analysis of social media text. In Proceedings of the international AAAI conference on web and social media, volume 8, pages 216–225. find
  20. J Peter Kincaid, Robert P Fishburne Jr, Richard L Rogers, and Brad S Chissom. 1975. Derivation of new readability formulas (automated readability index, fog count and flesch reading ease formula) for navy enlisted personnel. Technical report, Institute for Simulation and Training. find
  21. John Kirchenbauer, Jonas Geiping, Yuxin Wen, Jonathan Katz, Ian Miers, and Tom Goldstein. 2023a. A watermark for large language models. In International Conference on Machine Learning, pages 17061–17084. PMLR. find our card
  22. John Kirchenbauer, Jonas Geiping, Yuxin Wen, Manli Shu, Khalid Saifullah, Kezhi Kong, Kasun Fernando, Aniruddha Saha, Micah Goldblum, and Tom Goldstein. 2023b. On the reliability of watermarks for large language models. arXiv preprint arXiv:2306.04634. find
  23. Moshe Koppel, Jonathan Schler, and Shlomo Argamon. 2009. Computational methods in authorship attribution. J. Am. Soc. Inf. Sci. Technol., 60(1):9–26. find
  24. Tharindu Kumarage, Joshua Garland, Amrita Bhattacharjee, K Trapeznikov, S Ruston, and Huan Liu. 2023. Stylometric detection of ai-generated text in twitter timelines. arxiv 2023. arXiv preprint arXiv:2303.03697. — [arXiv:2303.03697](https://arxiv.org/abs/2303.03697) find
  25. Laida Kushnareva, Daniil Cherniavskii, Vladislav Mikhailov, Ekaterina Artemova, Serguei Barannikov, Alexander Bernstein, Irina Piontkovskaya, Dmitri Piontkovski, and Evgeny Burnaev. 2021. Artificial text detection via examining the topology of attention maps. arXiv preprint arXiv:2109.04825. find
  26. Aiwei Liu, Leyi Pan, Xuming Hu, Shu’ang Li, Lijie Wen, Irwin King, and Philip S Yu. 2023a. An unforgeable publicly verifiable watermark for large language models. arXiv preprint arXiv:2307.16230. — [arXiv:2307.16230](https://arxiv.org/abs/2307.16230) find
  27. Xiaoming Liu, Zhaohan Zhang, Yichen Wang, Hang Pu, Yu Lan, and Chao Shen. 2023b. Coco: Coherence-enhanced machine-generated text detection under low resource with contrastive learning. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 16167–16188. find
  28. Evan Lucas and Timothy Havens. 2023. Gpts don’t keep secrets: Searching for backdoor watermark triggers in autoregressive language models. In Proceedings of the 3rd Workshop on Trustworthy Natural Language Processing (TrustNLP 2023), pages 242–248. find
  29. Yong Luo, Huaizheng Zhang, Yongjie Wang, Yonggang Wen, and Xinwen Zhang. 2018. Resumenet: A learning-based framework for automatic resume quality assessment. In 2018 IEEE International Conference on Data Mining (ICDM), pages 307–316. IEEE. find
  30. Eric Mitchell, Yoonho Lee, Alexander Khazatsky, Christopher D Manning, and Chelsea Finn. 2023. Detectgpt: Zero-shot machine-generated text detection using probability curvature. In International conference on machine learning, pages 24950–24962. PMLR. find our card
  31. Abiodun Modupe, Turgay Celik, Vukosi Marivate, and Oludayo O Olugbara. 2022. Post-authorship attribution using regularized deep neural network. Applied Sciences, 12(15):7518. find
  32. Dena F Mujtaba and Nihar R Mahapatra. 2024. Fairness in ai-driven recruitment: Challenges, metrics, methods, and future directions. arXiv preprint arXiv:2405.19699. — [arXiv:2405.19699](https://arxiv.org/abs/2405.19699) find
  33. Travis Munyer, Abdullah All Tanvir, Arjon Das, and Xin Zhong. 2024. Deeptextmark: a deep learning-driven text watermarking approach for identifying large language model generated text. Ieee Access, 12:40508–40520. find
  34. Tempestt Neal, Kalaivani Sundararajan, Aneez Fatima, Yiming Yan, Yingfei Xiang, and Damon Woodard. 2017. Surveying stylometry techniques and applications. ACM Comput. Surv., 50(6). find
  35. Fabian Pedregosa, Gaël Varoquaux, Alexandre Gramfort, Vincent Michel, Bertrand Thirion, Olivier Grisel, Mathieu Blondel, Peter Prettenhofer, Ron Weiss, Vincent Dubourg, et al. 2011. Scikit-learn: Machine learning in python. the Journal of machine Learning research, 12:2825– 2830. find
  36. Lip Yee Por, KokSheik Wong, and Kok Onn Chee. 2012. Unispach: A text-based data hiding method using unicode space characters. Journal of Systems and Software, 85(5):1075–1082. find
  37. Martin Potthast, Sarah Braun, Tolga Buz, Fabian Duffhauss, Florian Friedrich, Jörg Marvin Gülzow, Jakob Köhler, Winfried Lötzsch, Fabian Müller, Maike Elisa Müller, Robert Paßmann, Bernhard Reinke, Lucas Rettenmeier, Thomas Rometsch, Timo Sommer, Michael Träger, Sebastian Wilhelm, Benno Stein, Efsta… find
  38. Shushanta Pudasaini, Subarna Shakya, Sagar Lamichhane, Sajjan Adhikari, Aakash Tamang, and Sujan Adhikari. 2021. Scoring of resume and job description using word2vec and matching them using gale–shapley algorithm. In Expert Clouds and Applications: Proceedings of ICOECA 2021, pages 705–713. Springer… find
  39. B Vamsi Nath Reddy and S Parthiban. 2025. Recognizing the required and skill-based resume to the companies with bert compared with random forest. In AIP Conference Proceedings, volume 3270, page 020155. AIP Publishing LLC. find
  40. Tal Schuster, Roei Schuster, Darsh J Shah, and Regina Barzilay. 2020. The limitations of stylometry for detecting machine-generated fake news. Computational Linguistics, 46(2):499–510. find
  41. Irene Solaiman, Miles Brundage, Jack Clark, Amanda Askell, Ariel Herbert-Voss, Jeff Wu, Alec Radford, Gretchen Krueger, Jong Wook Kim, Sarah Kreps, et al. 2019. Release strategies and the social impacts of language models. arXiv preprint arXiv:1908.09203. find our card
  42. Efstathios Stamatatos. 2009. A survey of modern authorship attribution methods. J. Assoc. Inf. Sci. Technol., 60:538–556. find
  43. Ruixiang Tang, Qizhang Feng, Ninghao Liu, Fan Yang, and Xia Hu. 2023. Did you train on my dataset? towards public dataset protection with cleanlabel backdoor watermarking. ACM SIGKDD Explorations Newsletter, 25(1):43–53. find
  44. Gemini Team, Rohan Anil, Sebastian Borgeaud, Jean-Baptiste Alayrac, Jiahui Yu, Radu Soricut, Johan Schalkwyk, Andrew M Dai, Anja Hauth, Katie Millican, et al. 2023. Gemini: a family of highly capable multimodal models. arXiv preprint arXiv:2312.11805. find our card
  45. Shikha Tewari, Dr Amit Joshi, and Deeksha Tewari. 2024. Ai powered hr: impact, benefits and challenges. BENEFITS AND CHALLENGES (October 17, 2024). find
  46. Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971. find
  47. KiYoon Yoo, Wonhyuk Ahn, Jiho Jang, and Nojun Kwak. 2023. Robust multi-bit natural language watermarking through invariant features. arXiv preprint arXiv:2305.01904. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/loizidou-2026-corpus-baselines-distinguishing-authentic-ai/