College Graduates’ Employment Recommendation using Hybrid Deep Learning based Convolutional Deep Semantic Structure Modelling

Employment recommendation for college graduates is highly crucial in the current job market to identify the correct career prospects. The proposed research aims at improving the functionality of job recommendation systems using Convolutional Deep Semantic Structure Model with Hybrid Deep Learning (CDSSM-HDL) model. The input is considered as the data from LinkedIn for in depth identification and processing of the semantic structures such as job postings and users’ profiles. The Hybrid Deep Learning (HDL) approach improves this feature extraction process by having multiple layers of neural networks from Multi-Layer Perceptron (MLP) and Artificial Neural Network (ANN) that learn higher-level features step by step. This resulted in providing better and more targeted recommendations to a particular job seeker or a particular job opening by reducing the sparsity and the cold-start issues inherent in recommendation systems. The experimental results showed that the proposed CDSSM-HDL model achieved better recommendation accuracy of 98.94%, recommended recall rate of 94.25%, and Mean Square Error (MSE) of 0.63, when compared to the existing methods such as Puppeteer and Representational State Transfer (REST) Application Programming Interface (API) and Artificial Neural Networks (ANN).