Personnel performance is important for the high-technology industry to ensure its core competitive advantages are present. Therefore, predicting personnel performance is an important research area in human resource management (HRM). In this paper, to improve prediction performance, we propose a novel framework for personnel performance prediction to help decision-makers to forecast future personnel performance and recruit the best suitable talents. Firstly, a hybrid convolutional recurrent neural network (CRNN) model based on self-attention mechanism is presented, which can automatically learn discriminative features and capture global contextual information from personnel performance data. Moreover, we treat the prediction problem as a classification task. Then, the k-nearest neighbor (KNN) classifier was used to predict personnel performance. The proposed framework is applied to a real case of personnel performance prediction. The experimental results demonstrate that the presented approach achieves significant performance improvement for personnel performance compared to existing methods.

Turnover prediction has an important impact on alleviating the brain drain, which can help organizations reduce costs and enhance competitiveness. Existing studies on turnover are mainly based on analyzing the turnover correlation, using different models to predict various employee turnover scenarios, and only predicting turnover category, while the class imbalance and turnover possibility have been ignored. To this end, in this paper, we propose a novel fine-grained adaptation-based turnover prediction neural network (FATPNN) model. Specifically, we first employ a GRU to learn profile-aware features representations of the personnel samples. Then, to evaluate the contribution of various turnover factors, we further exploit an attention mechanism to model the profile information. Finally, we creatively design a weighted-based probability loss function suitable for our turnover prediction tasks. Experimental results show the effectiveness and universality of the FATPNN model in terms of turnover prediction.

Online recruitment services are rapidly transforming traditional hiring practices in the job market. Accurate person-job fit is crucial for intelligent recruitment. Previous studies on person-job fit fail to explore job seekers’ resume information from a multi-perspective approach, and neglect the sustainable learning of resume features. To address this, the present paper proposes a Graph-based Person-Job Fit Neural Network Fusion (GPJFNNF) model. Specifically, the model first generates local semantic representations of job requirements and resume text using the BERT model. Next, a graph structure is constructed based on historical successful recruitment records, and the constructed resume graph is input into a graph neural network to obtain a global semantic representation of the resume. Finally, the adaptive feature fusion mechanism is used to fuse the local and global semantics of the resume, and the final semantic representation of the resume, along with the semantic representation of the job requirements, being input into the person-job fit layer. Experimental results demonstrate that the proposed model achieves 94.63%, 94.15%, 95.04%, and 94.59% in the person-job fit task in terms of accuracy, precision, recall, and F1, respectively, significantly outperforming state-of-the-art baselines.

Online recruitment platforms are transforming talent acquisition paradigms, where a precise person-job fit plays a pivotal role in intelligent recruitment systems. However, current methodologies predominantly rely on coarse-grained semantic analysis, failing to address the textual structural dependencies and noise inherent in resumes and job descriptions. To bridge this gap, the novel fine-grained semantics-enhanced graph neural network for person-job fit (FSEGNN-PJF) framework is proposed. First, graph topologies are constructed by modeling word co-occurrence relationships through pointwise mutual information and sliding windows, followed by graph attention networks to learn graph structural semantics. Second, to mitigate textual noise and focus on critical features, a differential transformer and self-attention mechanism are introduced to semantically encode resumes and job requirements. Then, a novel fine-grained semantic matching strategy is designed, using the enhanced feature fusion strategy to fuse the semantic features of resumes and job positions. Extensive experiments on real-world recruitment datasets demonstrate the effectiveness and robustness of FSEGNN-PJF.