Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Aug 25:8:1564521.
doi: 10.3389/fdata.2025.1564521. eCollection 2025.

Toward more realistic career path prediction: evaluation and methods

Elena Senger  1   2 , Yuri Campbell  2 , Rob van der Goot  3 , Barbara Plank  1
Affiliations

Toward more realistic career path prediction: evaluation and methods

Elena Senger et al. Front Big Data. .

Abstract

Predicting career trajectories is a complex yet impactful task, offering significant benefits for personalized career counseling, recruitment optimization, and workforce planning. However, effective career path prediction (CPP) modeling faces challenges including highly variable career trajectories, free-text resume data, and limited publicly available benchmark datasets. In this study, we present a comprehensive comparative evaluation of CPP models-linear projection, multilayer perceptron (MLP), LSTM, and large language models (LLMs)-across multiple input settings and two recently introduced public datasets. Our contributions are threefold: (1) we propose novel model variants, including an MLP extension and a standardized LLM approach, (2) we systematically evaluate model performance across input types (titles only vs. title+description, standardized vs. free-text), and (3) we investigate the role of synthetic data and fine-tuning strategies in addressing data scarcity and improving model generalization. Additionally, we provide a detailed qualitative analysis of prediction behaviors across industries, career lengths, and transitions. Our findings establish new baselines, reveal the trade-offs of different modeling strategies, and offer practical insights for deploying CPP systems in real-world settings.

Keywords: LLM; career path prediction; labor market; recommendation; synthetic data.

PubMed Disclaimer

Conflict of interest statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

Figure 1
Figure 1
Representation learning stage, adapted from (Decorte et al. 2023).
Figure 2
Figure 2
Metrics for DECORTE across ESCO categories. The numbers in the brackets indicate the number of data points available for the subset.
Figure 3
Figure 3
Metrics for DECORTE across subsets. The numbers in the brackets indicate the number of data points available for the subset.
Figure 4
Figure 4
Metrics for K+cp across ESCO categories. The numbers in the brackets indicate the number of data points available for the subset.
Figure 5
Figure 5
Metrics for K+cp across subsets. The numbers in the brackets indicate the number of data points available for the subset.
Figure 6
Figure 6
Metrics for KARRIEREWEGE+ datasets across subsets. The scores are the mean across all approaches for R@5. The numbers in the brackets indicate the number of data points available for the subset.
Figure 7
Figure 7
Normalized top-1 confusion matrices for the MLP approach across four datasets. Values represent row-wise proportions.
Figure 8
Figure 8
Absolute top-1 confusion matrices for the MLP approach across four datasets. Values represent raw prediction counts.
See this image and copyright information in PMC

References

    1. Bao K., Zhang J., Wang W., Zhang Y., Yang Z., Luo Y., et al. (2025). A Bi-Step Grounding Paradigm for Large Language Models in Recommendation Systems. New York, NY: Association for Computing Machinery. 10.1145/3716393 - DOI
    1. Cerilla M. T., Santillan A., Vinas C. J., Fuente M. B. D. (2023). “Career path modeling and recommendations with linkedin career data and predicted salary estimations,” in The First Tiny Papers Track at ICLR 2023, Tiny Papers @ ICLR 2023, eds. K. Maughan, R. Liu, and T. F. Burns (Kigali, Rwanda: ICLR).
    1. Chang L.-T., Simon L., Rajkumar K., Athey S. (2019). A Bayesian Approach to Predicting Occupational Transitions.
    1. de Groot M., Schutte J., Graus D. (2021). Job posting-enriched knowledge graph for skills-based matching. arXiv [preprint] arXiv:2109.02554. 10.48550/arXiv.2109.02554 - DOI
    1. Decorte J.-J., Hautte J. V., Deleu J., Develder C., Demeester T. (2023). “Career path prediction using resume representation learning and skill-based matching,” in RECSYS in HR 2023 : the 3rd Workshop on Recommender Systems for Human Resources (RecSys in HR 2023), eds. M. Kaya, T. Bogers, D. Graus, C. Johnson, and J. J. Decorte (Singapore: CEUR).

LinkOut - more resources