An Extractive-Abstractive Approach for Multi-document Summarization of Scientific Articles for Literature Review

Research in the biomedical domain is constantly challenged by its large amount of ever-evolving textual information. Biomedical researchers are usually required to conduct a literature review before any medical intervention to assess the effectiveness of the concerned research. However, the process is time-consuming, and therefore, automation to some extent would help reduce the accompanying information overload. Multi-document summarization of scientific articles for literature reviews is one approximation of such automation. Here in this paper, we describe our pipelined approach for the aforementioned task. We design a BERT-based extractive method followed by a BigBird PEGASUS-based abstractive pipeline for generating literature review summaries from the abstracts of biomedical trial reports as part of the Multi-document Summarization for Literature Review (MSLR) shared task¹ in the Scholarly Document Processing (SDP) workshop 2022². Our proposed model achieves the best performance on the MSLR-Cochrane leaderboard³ on majority of the evaluation metrics. Human scrutiny of our automatically generated summaries indicates that our approach is promising to yield readable multi-article summaries for conducting such literature reviews.