How Can Graph Neural Networks Help Document Retrieval: A Case Study on CORD19 with Concept Map Generation

TL;DR

This paper investigates how graph neural networks can enhance document retrieval by representing unstructured texts as concept maps, demonstrating that semantics-oriented GNNs outperform complex structure-oriented models on the CORD-19 dataset.

Contribution

The study introduces semantics-oriented graph functions for GNNs that improve document retrieval performance over traditional structure-oriented GNNs like GINs and GATs.

Findings

Semantics-oriented GNNs outperform structure-oriented GNNs in document retrieval.

The proposed methods achieve more stable performance based on BM25 candidates.

Insights guide future development of GNNs with semantic biases for textual reasoning.

Abstract

Graph neural networks (GNNs), as a group of powerful tools for representation learning on irregular data, have manifested superiority in various downstream tasks. With unstructured texts represented as concept maps, GNNs can be exploited for tasks like document retrieval. Intrigued by how can GNNs help document retrieval, we conduct an empirical study on a large-scale multi-discipline dataset CORD-19. Results show that instead of the complex structure-oriented GNNs such as GINs and GATs, our proposed semantics-oriented graph functions achieve better and more stable performance based on the BM25 retrieved candidates. Our insights in this case study can serve as a guideline for future work to develop effective GNNs with appropriate semantics-oriented inductive biases for textual reasoning tasks like document retrieval and classification. All code for this case study is available at https://github.com/HennyJie/GNN-DocRetrieval.