ConGraT: Self-Supervised Contrastive Pretraining for Joint Graph and Text Embeddings

TL;DR

ConGraT is a self-supervised pretraining method that jointly learns text and graph node representations, improving performance on multiple graph and text tasks by aligning embeddings in a shared space.

Contribution

The paper introduces ConGraT, a novel contrastive pretraining approach that jointly learns text and graph node embeddings without task-specific labels, incorporating graph structure into the contrastive objective.

Findings

Outperforms baselines on classification, link prediction, and language modeling.

Effectively incorporates graph structure into contrastive learning.

Enables community detection based on textual and structural information.

Abstract

Learning on text-attributed graphs (TAGs), in which nodes are associated with one or more texts, has been the subject of much recent work. However, most approaches tend to make strong assumptions about the downstream task of interest, are reliant on hand-labeled data, or fail to equally balance the importance of both text and graph representations. In this work, we propose Contrastive Graph-Text pretraining (ConGraT), a general, self-supervised approach for jointly learning separate representations of texts and nodes in a TAG. Our method trains a language model (LM) and a graph neural network (GNN) to align their representations in a common latent space using a batch-wise contrastive learning objective inspired by CLIP. We further propose an extension to the CLIP objective that leverages graph structure to incorporate information about inter-node similarity. Extensive experiments demonstrate that ConGraT outperforms baselines on various downstream tasks, including node and text category classification, link prediction, and language modeling. Finally, we present an application of our method to community detection in social graphs, which enables finding more textually grounded communities, rather than purely graph-based ones. Code and certain datasets are available at https://github.com/wwbrannon/congrat.