Fine-Grained Semantics-Enhanced Graph Neural Network Model for Person-Job Fit.

Area of Science:

• Artificial Intelligence
• Natural Language Processing
• Human-Computer Interaction

Background:

• Intelligent recruitment systems rely on accurate person-job fit.
• Current methods use coarse-grained semantic analysis, ignoring textual structure and noise in resumes/job descriptions.

Purpose of the Study:

• Propose a novel fine-grained semantics-enhanced graph neural network for person-job fit (FSEGNN-PJF).
• Improve the accuracy of talent acquisition by addressing limitations in current recruitment methodologies.

Main Methods:

• Construct graph topologies using word co-occurrence (pointwise mutual information, sliding windows).
• Employ graph attention networks for learning graph structural semantics.
• Utilize differential transformer and self-attention for semantic encoding of resumes and job requirements.
• Implement a fine-grained semantic matching strategy with enhanced feature fusion.

Main Results:

• Demonstrated effectiveness and robustness of the FSEGNN-PJF framework.
• Achieved superior performance in person-job fit assessment compared to existing approaches.
• Successfully mitigated textual noise and focused on critical features for better matching.

Conclusions:

• The FSEGNN-PJF framework significantly advances person-job fit analysis in intelligent recruitment.
• Fine-grained semantic analysis and graph neural networks offer a more robust approach to talent acquisition.
• This method provides a promising direction for optimizing online recruitment platforms.