An Attention-Based Deep Learning Framework for Intelligent Resume–Job Matching

Abstract

The increasing adoption of online recruitment platforms has significantly transformed modern human resource management by enabling large-scale and low-cost talent acquisition. However, the explosive growth of digital resumes and diverse job descriptions has created substantial challenges for recruiters in accurately and efficiently identifying suitable candidates. Manual screening procedures are often inefficient and prone to subjective bias. To address these limitations, this study develops an intelligent resume–job matching framework based on deep neural networks for semantic representation and relevance modeling. The proposed architecture integrates bidirectional long short-term memory networks and convolutional neural networks to capture both contextual dependencies and local lexical patterns from recruitment texts. Through joint feature learning, meaningful representations of candidate profiles and job requirements are constructed. Furthermore, an attention-based interaction mechanism is introduced to dynamically evaluate the relative importance between professional experiences and position specifications. This mechanism enables the model to emphasize key competency-related information and suppress irrelevant content during the matching process. By jointly exploiting contextual modeling, local feature extraction, and attention-driven weighting, the proposed approach achieves more accurate and robust alignment between resumes and job postings. Experimental results demonstrate that the method significantly improves matching precision and screening efficiency, providing effective support for intelligent recruitment systems.