Biomedical Text Mining: Biocuration and Literature Search Engines

Abstract

Many bioinformatics tasks, such as database development and computational predictions, rely heavily on thorough literature searches and reviews. Literature searches and reviews also form a foundation for research across domains - including bioinformatics, health sciences, agriculture and other life sciences. However, a literature review or (manual/automated) text mining will not be reliable without a meticulous collection of relevant articles. The proliferation of scientific journals and publications and an increasing focus on interdisciplinary research have made literature searches a complex and demanding task. This chapter begins with a quick review of the purpose of literature searches before discussing the information extraction methods, focusing on the manual curation process. We also briefly compare manual and automated data extraction approaches. A few insights are also shared from our own experiences, especially those related to building databases for gene expression and COVID-19. We highlight PubMed and Google Scholar as the most popular search engines for scientific literature, review their utilities, and advise caution when using them without a detailed feature analysis; we particularly point out some critical observations about potentially misleading PubMed query features. Next, we refer to a semi-quantitative comparative analysis of different search engines, which forms the most comprehensive comparison in the field to date, and examine the conclusions of this study in the context of eventual developments. We discuss other recent studies in the landscape of search engines. Given the rapidly evolving nature of search technologies, especially with the integration of artificial intelligence (AI), there is a pressing need for further comprehensive comparative studies to guide researchers in selecting the most suitable search engines.