Structured Multi-Label Biomedical Text Tagging via Attentive Neural Tree Decoding

TL;DR

This paper introduces an attentive neural tree decoding model for multi-label biomedical text tagging, effectively leveraging ontological structures to improve automatic MeSH term assignment from unstructured abstracts.

Contribution

It presents a novel sequence-to-sequence approach that recursively expands tree nodes based on input text, enhancing multi-label tagging accuracy in biomedical domain.

Findings

Outperforms state-of-the-art methods on MeSH term assignment

Utilizes tree-structured ontology for improved tagging

Demonstrates effectiveness in biomedical text annotation

Abstract

We propose a model for tagging unstructured texts with an arbitrary number of terms drawn from a tree-structured vocabulary (i.e., an ontology). We treat this as a special case of sequence-to-sequence learning in which the decoder begins at the root node of an ontological tree and recursively elects to expand child nodes as a function of the input text, the current node, and the latent decoder state. In our experiments the proposed method outperforms state-of-the-art approaches on the important task of automatically assigning MeSH terms to biomedical abstracts.