Deep Learning Framework Using HR-Net for Improving Recruitment and Retention Strategies in HR Management Systems
Abstract
HRM systems need to have intelligent analytical frameworks to aid in making good recruitment and staffing decisions. There is a tendency of existing systems to fail to analyze structured employee attributes and unstructured textual feedback in a joint way. In this paper, a hybrid neural network (HR-NET) will be introduced, which is an architecture that combines numerical HR data with semantic information obtained based on text records of employees. The framework uses a dual-path structure called Multi-Layer Perceptron, to structure features, and transformer-based natural language processing model to process unstructured data, which are merged with a fusion layer. The given framework is assessed based on standard classification metrics and compared to the traditional machine learning models. As it has been demonstrated, the experimental outcomes indicate that the HRNET has a total classification accuracy of 99.5 which makes it effective in recruitment and employee attrition prediction challenges. The results indicate that multimodal deep learning models have the potential to increase the use of data to make decisions in contemporary HR management systems.