HGCN4MeSH: Hybrid graph convolution network for MeSH indexing

Abstract

Recently, deep learning has been used in Medical Subject Headings (MeSH) indexing to reduce the labor costs associated with manual annotation, including DeepMeSH, TextCNN, etc. However, these models fail to capture the complex correlations between MeSH terms. To this end, we use a Graph Convolution Network (GCN) to learn the relationship between these terms and present a novel Hybrid Graph Convolution Net for MeSH index (HGCN4MeSH). We utilize two bidirectional GRUs to learn the embedding representation of the abstract and the title of the MeSH index text respectively. We construct the adjacency matrix of MeSH terms, based on the co-occurence relationships in corpus, and use the matrix to learn representations using the GCN. On the basis of learning the joint representation, the prediction problem of the MeSH index keywords is an extreme multi-label classification problem after the attention layer operation. Experimental results on two datasets show that HGCN4MeSH is competitive with the state-of-the-art methods.