Page 148 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 43

148 Journal of Library Science in China, Vol.9, 2017

count of the article is 0, but its usage count reaches as high as 296, ranking the article among the
top four of all Library and Information Science articles. A higher usage count can also constitute
the influence of the literature and the author. In specific cases, usage has indeed manifested its
enormous potential in supplementing citation data and index, thereby providing richer evaluation
perspectives.

4 Conclusions

This research first takes a theoretical perspective to discuss the potential value and basic nature
of the usage data of the WoS platform. Then, using 166,767 articles in physics, computer science,
economics, and Library and Information Science as research samples and original data, this
research empirically studies the basic statistical features and bibliometric distribution model of
the magnitude of data usage for academic literature, and compares usage data with citation data
in terms of value and ranking. Finally, the paper introduces the representative literature of the
discipline of Library and Information Science as cases for a qualitative discussion. According
to the results, compared with citation data, usage is more discriminative and sensitive. In terms
of distribution, usage presents an approximate and positively skewed distribution at the high-
frequency part whilst an approximate power-law distribution is observed in a cumulative
integral. When the number of articles is divided into three equal portions, the cumulative usage
count of each portion satisfies the approximate relation, n :n:1. The evaluation result of usage
2
is independent to a certain degree, and there is no essential opposition to the result of citation.
These results suggest that usage can serve as a type of supplementary data for academic
evaluation.
Although the usage of the WoS platform can be expected to become an important data
source for future research on academic literature usage, it certainly has its own limitations as a
characterization of the academic influence of one aspect. First, the usage data consist of the full-
text download and bibliographic data export count, which does have its novelty, but the mixing of
the two data inevitably leads to confusions about their overall physical meaning or the generality of
meaning. Second, as the data generated by platform users, usage has a relatively high falsifiability.
Taking the citation of articles included by WoS as an example, the premise of falsifying a citation
is to publish at least one article included by SCI/SSCI/AHCI. However, the change in usage
value has no such premise and can be realized by any user with access authority to this platform.
If the falsifiability problem cannot be properly solved, the index will become susceptible to the
paradox of “failed as soon as becoming the standard”. The third aspect involves the relatively
high isolation degree of the data. Currently, WoS only publishes the usage data from the date of
the publication of the article to the date of retrieval, and within the last six months, but WoS does
not provide details about usage, so the usage data are basically a black box-type of data. As far
as openness is concerned, usage is inferior to citation data, which support the query of the source