Intelligent College Employment Recommendation Fusing Multimodal Sentiment Analysis with Knowledge Graph Reasoning

To overcome the dual obstacles of emotional blindness and misaligned labor market matching in college career counseling, we design an intelligent assistance system that synergistically combines multimodal sentiment analytics with an employment-oriented knowledge graph. Technically, we propose a novel Cross-modal Attention Fusion Network (CAFN) that, for the first time, employs a cross-modal co-attention mechanism coupled with sequential sentiment evolution tracking to synthesize information from in-class videos and interview transcripts, achieving state-of-the-art accuracy in uncovering latent career-related psychological distress. Simultaneously, we construct a semantically enriched three-layer employment knowledge graph and apply a GNN-based hierarchical attention propagation algorithm to deliver fine-grained and transparent recommendation decisions. Evaluated on a sample of 360 students from West Anhui University, our approach demonstrates substantial practical value: job-fit satisfaction surges by 30.3%, and the proportion of students transitioning from intention to actual employment climbs by 37.5%. Collectively, the proposed framework furnishes an objective, reproducible, and AI-infused methodology that directly confronts the “employment difficulty” issue, offering a robust reference for other institutions facing similar challenges.