BIT.UA at Biocreative VIII track 1: A joint model for relation classification and novelty detection

Abstract

The task of relation extraction has long posed a challenge within the Natural Language Processing (NLP) community, and its application in biomedical research is important for understanding scientific literature. The development of a tool capable of effectively addressing this task holds the potential to improve knowledge discovery by automating the extraction of relations from literature. The first track in the Biocreative VIII competition extended the scope of this challenge by introducing the detection of novel relations within literature. This paper presents the strategies used in this competition by our team, Biomedical Informatics and Technologies (BIT) at the University of Aveiro. We leveraged joint training to craft a singular, versatile model capable of not only classifying relations between two entities but also determining the novelty of the identified relation. Our experiments yielded promising results, with our submission outperforming the competition's average. This paper not only details our approach but also highlights the potential of joint training in relation extraction, paving the way for improved automated analysis of biomedical literature.

This article is part of the Proceedings of the BioCreative VIII Challenge and Workshop: Curation and Evaluation in the era of Generative Models.