MIReAD: Simple Method for Learning High-quality Representations from Scientific Documents
This addresses the need for better academic literature search and recommendation systems, though it is incremental as it builds on existing transformer methods.
The authors tackled the problem of learning high-quality document-level representations for scientific articles, proposing MIReAD, a method that fine-tunes a transformer model to predict journal classes from abstracts, and showed it outperforms six existing models across four evaluation standards.
Learning semantically meaningful representations from scientific documents can facilitate academic literature search and improve performance of recommendation systems. Pre-trained language models have been shown to learn rich textual representations, yet they cannot provide powerful document-level representations for scientific articles. We propose MIReAD, a simple method that learns high-quality representations of scientific papers by fine-tuning transformer model to predict the target journal class based on the abstract. We train MIReAD on more than 500,000 PubMed and arXiv abstracts across over 2,000 journal classes. We show that MIReAD produces representations that can be used for similar papers retrieval, topic categorization and literature search. Our proposed approach outperforms six existing models for representation learning on scientific documents across four evaluation standards.