Paper Number
ICIS2025-2264
Paper Type
Short
Abstract
Predicting talent potential remains a fundamental challenge in workforce analytics. Traditional point-in-time assessments fail to capture the dynamic nature of professional growth across organizations and time. We propose LADDER-GNN (Level-Aware Dynamic Development \& Entity Representation GNN), a novel framework that explicitly incorporates hierarchical information across entities. Our model addresses key limitations in existing approaches by capturing both entity transitions and movements across career ladders while accounting for parallel experiences and gap periods. Through a two-stage learning process combining contrastive learning and joint task optimization, LADDER-GNN outperforms state-of-the-art methods in predicting next job position, next employer, and employment duration. We introduce a Career Projection Space mapping talent across "Ability to Thrive" and "Willingness to Stay" dimensions, enabling organizations to develop targeted talent management strategies. Extensive experiments on a large-scale OPN dataset demonstrate LADDER-GNN’s superior predictive accuracy and enhanced interpretability for strategic human resource management.
Recommended Citation
Xu, Dawei; Yang, Jingyuan; Zhong, Hao; and Li, Xitong, "Who Thrives, Who Stays? Predicting Talent Potential from Level-Aware Career Moves" (2025). ICIS 2025 Proceedings. 12.
https://aisel.aisnet.org/icis2025/da_bus/da_bus/12
Who Thrives, Who Stays? Predicting Talent Potential from Level-Aware Career Moves
Predicting talent potential remains a fundamental challenge in workforce analytics. Traditional point-in-time assessments fail to capture the dynamic nature of professional growth across organizations and time. We propose LADDER-GNN (Level-Aware Dynamic Development \& Entity Representation GNN), a novel framework that explicitly incorporates hierarchical information across entities. Our model addresses key limitations in existing approaches by capturing both entity transitions and movements across career ladders while accounting for parallel experiences and gap periods. Through a two-stage learning process combining contrastive learning and joint task optimization, LADDER-GNN outperforms state-of-the-art methods in predicting next job position, next employer, and employment duration. We introduce a Career Projection Space mapping talent across "Ability to Thrive" and "Willingness to Stay" dimensions, enabling organizations to develop targeted talent management strategies. Extensive experiments on a large-scale OPN dataset demonstrate LADDER-GNN’s superior predictive accuracy and enhanced interpretability for strategic human resource management.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.
Comments
07-DataAnalytics