Predicting cross-domain collaboration using multi-task learning

Zhenyu Hu, Jingya Zhou, Wenqi Wei, Congcong Zhang, Yingdan Shi

Abstract: Highlights•We propose a new method to capture the motivations for cross-domain collaboration.•We define the collaborative pattern to model the complex topic-topic interactions.•We design a multi-task framework CDCP to boost collaboration prediction performance.•CDCP shows excellent prediction performance on large publication datasets.