Evaluating Research Trends Using Key Term Occurrences and Multivariate Mann-Kendall Test

This paper offers a quantifiable means for assessing the trend in the interest of scientific community in a particular research theme. Representing the research theme as a set of key terms, we investigate its trend within the scientific literature, using a corpus of paper metadata belonging to flagship journals and conferences that are relevant for the encompassing domain. Our method employs a multivariate version of the Mann-Kendall test that is applied to a multivariate time series of key term occurrences in order to discover if the trend of the research theme is ascending, descending or if there is no monotonic trend. Finally, an illustrative case study, examining the effectiveness of the proposed method for three research themes from Electronic Design Automation field, is presented.