FQ-GNN: FIREFLY-OPTIMIZED QUANTUM GRAPH NEURAL NETWORKS FOR CONTEXTUAL DEPRESSION DETECTION FROM SOCIAL MEDIA TEXTS

Abstract

FQ-GNN is aphase of research work that presents a novel framework related to depression detection from social media textswith the use of a quantum graph neural network improvedthrough a bio-inspired Firefly algorithm. This proposed approach constructs a heterogeneous graph where social media users, posts, and interactions are modeled as edges and nodes weighted bytemporal decay and semantic similarity. Node-level embeddings are derived through a transformer-based semantic encoder integrated with temporal context filtering, catching both interaction chronology and textual content. Then, these embeddings are managed through quantum-aware graph convolutional layers to extract rich semantic and structural features. To professionally explore the high-dimensional parameter space and maximizethe multi-objective performance metrics, a firefly-driven quantum hyper-parameter optimizer is employed. Lastly, adaptive fusion and neighbor-aware refinement combine structural and semantic information to make accurate depression calculations. Extensive evaluations establish that FQ-GNN significantly outperforms conventional baselines, providing contextually-aware detection of depressive behavior in online social interactions and robustness.