Automatic Matching of Processes and Employees Based on Transformer and Multi-level Understanding

When implementing processes, power companies need to identify process risks and assign suitable employees to complete the corresponding tasks. Traditional process matching strategies rely on manual matching of employees and processes, which results in high labor costs. With the advancement of artificial intelligence, automated matching of processes and employees based on AI has become possible. In this paper, we propose a novel method for automatic matching of processes and employees based on Transformer and multi-level understanding. This method first uses a text encoder to extract embedding vectors from the descriptions related to employees and processes. Then, we design a multi-level understanding module to help the model further focus on risk-related semantic information, improving the quality of the embedding vectors. Finally, we use Transformer to obtain the CLS embedding vectors corresponding to employees and processes, and calculate similarity using these vectors to achieve precise matching. Experimental results on a proprietary dataset demonstrate the effectiveness of our method.