Job seekers aim to find job offers that align with their qualifications, as do human resource departments in seeking candidates whose resumes match their expectations. In this work, the proposal suggests using graph representation for data and treating the problem as a classification task. The proposed DGL4C model, a semi-supervised graph deep learning approach, learns representations from graphs and trains a classifier on this latent data. Experiments conducted on an anonymous resume dataset demonstrate that DGL4C notably enhances precision and accuracy compared to traditional deep learning models.