Use of relative extra citation counts and uncited publications to enhance the discriminatory power of the h-index

Abstract

Although the conventional h-index attempts to combine a researcher’s publications and citation-based impact, it requires threshold citation counts and cited papers to award a score. The consequence is that the h-index fails to consider extra citation counts as well as uncited publications, both of which reveal a researcher’s performance output. Recently, an apparent h-index was presented to extend the conventional h-index to account for the uncited publications of a researcher; unfortunately, the apparent h-index also fails to account for a researcher’s extra citation counts. In this paper, we extend the apparent h-index to account for a researcher’s extra citation counts. In particular, contrary to the conventional h-index and its variants, the proposed author-level metric uses relative extra citation counts and uncited publications to discriminate among researchers. The relative extra citation count is defined as the ratio of the absolute extra citation counts to the total number of citations. The proposed index, called a comprehensive h-index (c-index), is formulated by incorporating the relative extra citation counts and the fraction of the cited publications into the h-index. The advantages of the c-index are two-folds: first, it enhances the discriminatory power of the h-index as it considers researcher’s entire publications, whether cited or not, and all the citation counts; and second, it preserves the desirable features of the conventional h-index, namely robustness and simplicity. Case studies for researchers in Chemistry, Physics, Material Science, Engineering and Medicine have been presented. Results show that researchers could have equal h-index, but unequal c-index due to their unequal relative extra citation counts, as well as their uncited publications. Furthermore, a comparison of the c-index with the h- and g-indices have been made; results show that the c-index is more discriminatory than both the h- and g-indices. The proposed metric may contribute to the ongoing discussions on the improvement of the conventional h-index.