Examining the characteristics of impactful research topics: A case of three decades of HIV-AIDS research

Abstract While research is a process that may span over several years, and its success depends on multiple factors such as performance of the scientists, methods and technologies employed, one cannot understate the importance of the choice of research topic. Yet, how do we systematically characterize a research topic, and to what extent do topic choices affect the future impact of scientific studies? Drawing on more than 200,000 HIV-AIDS-related publications over the last three decades, we constructed a dynamic topic network from the abstracts of these publications through a topic-modeling approach. The network depicts the evolution of HIV-AIDS scientific knowledge over time. We mapped some salient network properties of each paper’s topic(s) to the future citation impact of the publication. Results showed that high-impact studies include novel topics (topics less-studied before, in contrast to “popular” topics), combine topics with varying maturity (e.g., well-explored and less-studied topics), and have a smaller number of competing topics. Furthermore, an excessive combination of topics has a detrimental effect on the impact of a publication.