Assessment of author ranking indices based on multi-authorship

Abstract

Various multi-authorship indices have been proposed in the literature, however, there is a continuous discussion in the scientific community which multi-authorship index performs better for the ranking of authors. So far, multi-authorship indices are assessed on very small datasets mostly single authors or the publication record of less than 10 authors. Furthermore, the indices are evaluated on different datasets, making it difficult to assess the true contribution and importance of each multi-authorship index over the others. To identify the individual performance of each multi-authorship index, we employ a comprehensive dataset of Civil Engineering domain, rank the authors according to each index and calculate the correlation between the ranked lists obtained by these indices. It is found that the correlation values vary between very strong correlation and very weak correlation and the negative correlation also exists between some of the indices. Secondly, to evaluate the ranking, the occurrence of award winners is found in the author ranked lists of these indices. The award winners of four most renowned societies of Civil Engineering were considered as benchmark. In top 10% of the ranked list, gf-index remained successful in bringing most of the awardees i.e. around 67% of total awardees. Overall, none of the multi-authorship index remained successful in bringing 100% of award winners in the list of top ranked authors.