Comparative Analysis of the Algorithm for Calculating the h-Index and Its Modifications

Introduction. The article presents a comparative analysis of existing algorithms for calculating the h-index and a number of its modifications that use the same basic idea or make use of its algorithm as the starting point. What makes the study relevant is the significant demand for a better approach to assessing the publication activity of researchers over the existing h-index algorithm. Methods. The study looks at the following 7 indexes that assess a researcher's scientific output: h-index; g-index (by Leo Egghe); j-index (the Mikhailov index); gh-index; hp-index; ghp-index; w-index. The main method used in the study is analysis; other formal logic methods that were also used in the study include synthesis, classification, deduction, and induction. Results and Discussion. The quantitative and qualitative indicators proposed in the article were used to carry out the compar-ative analysis of the algorithms behind all of the aforementioned indexes. One common feature of all these algorithms is that they all rely on the basic h-index procedure, while the difference is the degree to which each index takes into account all the publications and citations of a specific researcher. It was established that only the algorithms used in the gh-, hp-, and ghp-indexes take into account the entire set of a given author's publications. Meanwhile, the gh-index should best be regarded as an index of basic publications (it establishes the significance of the researcher's publications), while the hp-index is best viewed as a measure of the intensity of the researcher's work. As far as the ghp-index is concerned, it should only be used in situations in which the gh- or hp-indexes produce the same results. Amongst the algorithms for all the indexes reviewed in the article, the ones that best reflect a researcher's achievements are the gh-, hp-, and ghp-indexes, but they are somewhat more complex than the h-, g- и j-indexes. Conclusion. The analysis we carried out should make it possible to select the best algorithm depending on the complexity of the distribution of citations of published papers in order to achieve the desired level of assessment of a researcher's publications.