EMPLOYMENT PREDICTION ALGORITHM BASED ON ATTENTION MECHANISM WITH BIDIRECTIONAL SELECTION GRAPH
Based on the potential bidirectional selection relationship between employees and companies,we can predict where the employees will go in the future and the prospective hiring situation of the company.These predictions are of great significance for practitioners in finance,recruitment and other industries to judge the future development of company,find potential candidates,etc.To obtain the bidirectional selection information and reinforce the vertex information to enhance the prediction effect,the companies,employees and their related data were used to form a heterogeneous information network(HIN),and the bidirectional selection embeddings and the node embeddings for employment prediction were generated by autoencoder and tri-attention mechanism.According to the experiments on real dataset,our model has a better effect on prediction compared with the existing methods.The interpretability of the prediction is also improved.
Heterogeneous information networkMeta-pathAutoencoderAttention mechanismDeep neural networkCareer mobility behavior
国家科技期刊平台
NSTL
万方数据
