Civil Servant Position Recommendation Algorithm Based on Model Fusion

Recommending positions for civil servants efficiently, scientifically, and accurately has always been an urgent need for government departments in personnel management. However, most of the existing position recommendation algorithms do not use the sequence information in the historical working experience of civil servants; and only use collaborative filtering or features for the recommendation. Our motivation is to solve the shortcomings of existing position recommendation algorithms and realize efficient and accurate recommendations. Therefore, we propose a model fusion recommendation algorithm based on sequence, effectively fusing sequence and feature information. We use SR-GNN (Session-based Recommendation with Graph Neural Networks) as our base learner to learn sequence information, and we also use logistic regression, Xgboost, LightGBM as our base learners to learn feature information. Finally, Xgboost is used as our meta learner to fuse the output of our base learner. Experiments on the open civil servant resume data set show that our method has achieved better results than the original sub learner.