Understanding the Usability of a Literature-Based Discovery System Among Clinical Researchers in Sarawak, Malaysia

Introduction

With the rapid growth of scientific publications, it is hard for researchers to keep track of the latest developments in their field of interest and to perform literature-based knowledge discovery for exploration of new research directions (Simpson and Demner-Fushman, 2012; Pyysalo et al., 2019, p. 1553). This consequently restricts the ability of researchers to generate effective and high priority research hypotheses from literature review. As reported earlier, up to 85% of biomedical research funding is wasted (Chalmers and Glasziou, 2009, p. 1344; Crowe et al., 2015; Moher et al., 2016, p. 1577) on addressing low priority or existing issues. This scenario reflects the uncertainties that arose from the process of research hypothesis formulation. A key step to generate a worthwhile research hypothesis is to support researchers in refining their initial research ideas by enabling them to discover new knowledge from large collections of literature (Moher et al., 2016, p. 1587). As a result, numerous Literature-Based Discovery (LBD) systems such as Arrowsmith, Bitola, Spark and LION LBD (Gopalakrishnan et al., 2019) have been developed to assist medical researchers in discovering new knowledge from bibliographic databases for hypothesis generation and testing (Sebastian et al., 2017). LBD is a systematic computational approach that seeks to reveal latent connections by bridging fragments of information from disjoint literatures using innovative technologies such as natural language processing, artificial intelligence, and information retrieval (Swanson, 1986, p. 7; Swanson and Smalheiser, 1997, p. 184; Swanson, 2008, p. 5; Smalheiser, 2011, p. 218; Sebastian et al., 2017; Gopalakrishnan et al., 2019; Pyysalo et al., 2019, p. 1554). The idea of LBD was first employed in medical fields to discover a drug as a treatment for a disease or a gene as the cause of a disease by linking two seemingly unrelated concepts together with a third concept from bibliographic databases (Swanson, 1986). The emergence of LBD systems enables researchers to discover new knowledge automatically from vast collection of literatures. The systems allow researchers to understand the relations between medical concepts and discover implicit links between literatures even though they are seemingly unrelated towards one another (Smalheiser, 2017, p. 51). This in turn enables them to discover useful information from existing literatures and direct them to generate empirically testable research hypotheses.