Classification of Research Papers on Radio Frequency Electromagnetic Field (RF-EMF) Using Graph Neural Networks (GNN)

This study compares the performance of graph convolutional neural network (GCN) models with conventional natural language processing (NLP) for classifying scientific literature related to radio frequency electromagnetic field (RF-EMF). Specifically, examines two GCN models: BertGCN and citation-based GCN. The concludes that model achieves consistently good when input text is long enough, based on attention mechanism BERT. When sequence short, composition parameter λ, which combines output values subnetworks BertGCN, plays a crucial role in achieving high classification accuracy. As value λ increases, accuracy also increases. proposes tests simplified variant revealing differences among under different data conditions by existence keywords. has main contributions: (1) implementation testing document tasks fields publications, (2) confirmation impact conditions, such as keywords length, original BertGCN. Although this focused specific domain, our approaches have broader implications extend beyond publications general classification.