Page 143 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 43

ZHAO Xing / Exploring the measurement features of usage data for academic literature 143

Combining the above-mentioned approximate and positively skewed distribution at the high-
frequency part and the approximate power-law distribution in a cumulative integral here, the
usage value distribution of the WoS platform has unique features, but has not completely broken
away from the basic theoretical paradigm of bibliometrics, and has maintained its novelty while
embodying a theoretical uniformity.

Figure 3. Distribution fitting of the cumulative number and cumulative usage count of articles in four disciplines.

3.3 Comparison of the measurement results of usage and citation

As relatively independent academic influence evaluation data, usage is supposed to be more
discriminative than citation in results. According to the data of this paper, the literature of the
four sample disciplines indeed presented some differences in the two indexes of citation count
and usage count. For instance, the physics article ranked in the fourth place in usage count has a
citation count of 0 whilst the economics article with the highest usage count has a citation count
of only 2. There are also some zero-citation articles in computer science and in Library and
Information Science, whose usage counts nevertheless come out the highest. These examples all
confirm the hypothesis of Inference 2-2. In terms of overall values, the usage count and citation