Article: Employment trajectory prediction for graduates using temporal graph convolutional networks Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.52 pp.81 - 101 Abstract: Addressing psychological and social factors in graduate employment prediction, this paper proposes a graph network model that integrates psychological time-series data with dynamic social relationships. Traditional methods use static academic data and cannot capture key psychological factors like anxiety and career efficacy, or their interaction with peer and alumni resources. By constructing a time-series graph from psychological scales and social ties, tested on public graduate data, the model achieves an area under the curve of 0.891 for employment prediction. It significantly outperforms long short-term memory networks (area under the curve 0.801) and static graph neural networks (area under the curve 0.832), with normalised discounted cumulative gain at rank position 5 of 0.882, demonstrating reliable destination ranking. This work provides a data-driven approach for precise employment guidance through psychological monitoring. Inderscience Publishers - linking academia, business and industry through research

Title: Employment trajectory prediction for graduates using temporal graph convolutional networks

Authors: Chengcheng Liu

Addresses: Department of Visual Arts, Hunan Mass Media Vocational and Technical College, Hunan, 410000, China

Abstract: Addressing psychological and social factors in graduate employment prediction, this paper proposes a graph network model that integrates psychological time-series data with dynamic social relationships. Traditional methods use static academic data and cannot capture key psychological factors like anxiety and career efficacy, or their interaction with peer and alumni resources. By constructing a time-series graph from psychological scales and social ties, tested on public graduate data, the model achieves an area under the curve of 0.891 for employment prediction. It significantly outperforms long short-term memory networks (area under the curve 0.801) and static graph neural networks (area under the curve 0.832), with normalised discounted cumulative gain at rank position 5 of 0.882, demonstrating reliable destination ranking. This work provides a data-driven approach for precise employment guidance through psychological monitoring.

Keywords: temporal graph convolutional network; T-GCN; employment trajectory prediction; mental health; dynamic social network.

DOI: 10.1504/IJICT.2026.153712

International Journal of Information and Communication Technology, 2026 Vol.27 No.52, pp.81 - 101

Received: 08 Jan 2026
Accepted: 09 Feb 2026
Published online: 21 May 2026 *

Title: Employment trajectory prediction for graduates using temporal graph convolutional networks

Keep up-to-date