Query Focused Multi-document Summarisation of Biomedical Texts

TL;DR

This paper describes a system for query-focused multi-document summarization of biomedical texts, utilizing BERT-based models, with experiments showing BERT + LSTM outperforms other variants.

Contribution

It introduces a framework applying classification or regression to sentence embeddings for biomedical summarization, exploring BERT, BioBERT, Siamese architectures, and reinforcement learning.

Findings

BERT with LSTM yields best summarization results.

Siamese architectures and BioBERT did not improve performance.

Reinforcement learning variants were explored but not superior.

Abstract

This paper presents the participation of Macquarie University and the Australian National University for Task B Phase B of the 2020 BioASQ Challenge (BioASQ8b). Our overall framework implements Query focused multi-document extractive summarisation by applying either a classification or a regression layer to the candidate sentence embeddings and to the comparison between the question and sentence embeddings. We experiment with variants using BERT and BioBERT, Siamese architectures, and reinforcement learning. We observe the best results when BERT is used to obtain the word embeddings, followed by an LSTM layer to obtain sentence embeddings. Variants using Siamese architectures or BioBERT did not improve the results.