Author:

Keywords:

Science & Technology, Technology, Computer Science, Interdisciplinary Applications, Computer Science, SCIENTIFIC LITERATURE, PERFORMANCE, JOURNALS

Abstract:

One of the most important requirements of building applicable models and meaningful indicators for the use of scientometrics at the micro and meso level is the correct identification and disambiguation of authors and institutes. Platforms like ResearcherID or ORCID with author registration providing high reliability but lower coverage now provide appropriate data sets for the development and testing of stochastic models describing the publication activity and citation impact of individual authors. This paper proposes a triangular model incorporating papers, citations and authors analogously to the dichotomous model used at higher levels of aggregation like countries or fields. This model is applied to a set of authors in any field of science identified by their ResearcherID. However, the main advantage of classical citation indicators to study citation impact under conditional productivity turned out to be the main problem in this triangle: the possible heterogeneity of the collaborating authors results in low robustness. A mere technical solution to this problem would be fractional counting at three levels, but the conceptual issue, the different roles of co-authors causing this heterogeneity, will never be solved by any algorithm.