PTCR-PJF: A Person-Job Fit Model for Structured Resumes

With the declining efficiency of online recruitment and the increasing trend of structured job seeker information, existing Person-job Fit (PJF) methods are challenged by insufficient matching accuracy and the complexity of data structures. Traditional models often fail to utilize information in structured resumes fully and are inefficient in processing large volumes of heterogeneous data. To address these issues, we propose a novel Person-Job Fit model (PTCR-PJF) based on an improved Text Convolutional Neural Network (TextCNN) and Roformer joint modeling. PTCR-PJF employs two position-aware strategies: sine-cosine absolute position encoding (SPE) and rotary relative position encoding (RoPE), effectively enhancing the recognition of specific positional information in job postings. Additionally, the model employs a dual-layer text convolution module and applies chunk max pooling and adaptive max pooling to extend its perception of positional information further. It also integrates a Roformer based on RoPE to capture long-distance dependencies. Extensive experiments conducted on the large-scale real-world dataset CareerBuilder2012 and comparisons with 12 baseline models show that PTCR-PJF achieves state-of-the-art (SOTA) performance. The results indicate a significant advantage of the PTCR-PJF model in processing structured resume data, offering a new technological path for online intelligent recruitment.