Patton: Language Model Pretraining on Text-Rich Networks

TL;DR

Patton is a novel pretraining framework for text-rich networks that integrates textual and structural information, significantly improving performance on various downstream tasks across multiple domains.

Contribution

This paper introduces Patton, the first pretraining method that combines network structure and text for improved modeling of text-rich networks.

Findings

Patton outperforms baseline models on four downstream tasks.

The framework is effective across academic and e-commerce datasets.

Incorporating network structure enhances language model performance.

Abstract

A real-world text corpus sometimes comprises not only text documents but also semantic links between them (e.g., academic papers in a bibliographic network are linked by citations and co-authorships). Text documents and semantic connections form a text-rich network, which empowers a wide range of downstream tasks such as classification and retrieval. However, pretraining methods for such structures are still lacking, making it difficult to build one generic model that can be adapted to various tasks on text-rich networks. Current pretraining objectives, such as masked language modeling, purely model texts and do not take inter-document structure information into consideration. To this end, we propose our PretrAining on TexT-Rich NetwOrk framework Patton. Patton includes two pretraining strategies: network-contextualized masked language modeling and masked node prediction, to capture the inherent dependency between textual attributes and network structure. We conduct experiments on four downstream tasks in five datasets from both academic and e-commerce domains, where Patton outperforms baselines significantly and consistently.