Experimenting with the partnership ability φ-index on a million computer scientists
Schubert introduced the partnership ability φ-index relying on a researcher’s number of co-authors and collaboration rate. As a Hirsch-type index, φ was expected to be consistent with Schubert–Glänzel’s model of h-index. Schubert demonstrated this relationship with the 34 awardees of the Hevesy medal in the field of nuclear and radiochemistry (r 2 = 0.8484). In this paper, we upscale this study by testing the φ-index on a million researchers in computer science. We found that the Schubert–Glänzel’s model correlates with the million empirical φ values (r 2 = 0.8695). In addition, machine learning through symbolic regression produces models whose accuracy does not exceed a 6.1 % gain (r 2 = 0.9227). These results suggest that the Schubert–Glänzel’s model of φ-index is accurate and robust on the domain-wide bibliographic dataset of computer science.
KeywordsPartnership ability index Co-authorship Empirical validation Symbolic regression
- Chen, J., & Konstan, J. A. (2010). Conference paper selectivity and impact. Communications of the ACM, 53(6), 79–83. doi: 10.1145/1743546.1743569.
- Hirsch, JE. (2005). An index to quantify an individual’s scientific research output. Proceedings of the National Academy of Sciences of the USA, 102(46), 16569–16572. doi: 10.1073/pnas.0507655102.
- Ley, M. (2002). The DBLP computer science bibliography: evolution, research issues, perspectives. In: A. H. F. Laender, A. L. Oliveira (eds.) SPIRE’02 : Proceedings of the 9th international conference on string processing and information retrieval (vol. 2476, pp. 1–10). Springer, LNCS. doi: 10.1007/3-540-45735-6_1.
- Lotka, A. J. (1926). The frequency distribution of scientific productivity. Journal of the Washington Academy of Sciences, 16(12), 317–324.Google Scholar