Page 67 - JOURNAL OF LIBRARY SCIENCE IN CHINA 2018 Vol. 44

066 Journal of Library Science in China, Vol.10, 2018

perspective of crowdsourcing management, it is also more complicated in business process and
service mode. Additionally, although the previous open source software (OSS) collaboration model
also embodies the concept and means of crowdsourcing to a large extent, and carries out in-depth
exploration of collaboration issues, the operation of OSS project mainly relies on the participation
of technical enthusiasts with good programming basis, which is not consistent with the idea that
the general public is expected to be the participant in scientific crowdsourcing activities, because
the latter often requires systematic planning and design on task descriptions, participant motivation
and volunteer training. Finally, compared with the traditional business crowdsourcing, the
scientific crowdsourcing activities generate a large amount of research data, which is richer, more
diverse and more complex with respect to data magnitude, data structure and type, and data value
density. Therefore, scientific crowdsourcing has higher requirements on data monitoring, data
quality management and knowledge discovery.
As for the classification of scientific crowdsourcing activities, there are few prior work paying
attention to this field. Based on the systematic theory, Geiger and Schader (2014) interpret the
classification from two dimensions according to the meta characteristics of crowdsourcing
activities, including homogeneous or heterogenous contributions, and emergent value or non-
emergent value. The authors assert that homogeneity and heterogeneity have certain difficulty and
ambiguity in the actual classification of scientific crowdsourcing activities, and the main problem
is how to determine the type and granularity of contributions. For instance, even though scientific
crowdsourcing tasks that participants are required to upload eventually is text content, uploading a
few keywords and uploading an entire paragraph are clearly not a work granularity. Similarly, for
image resources, figure, picture and image also differ in contribution granularity. In view of this,
firstly, from the angle of the adoption of scientific crowdsourcing activities, this paper, based on the
user feedback results collected widely, considers whether it has emergent property. Non-emergent
scientific crowdsourcing refers to the fact that every individual’s contribution is complete and is
unable or unnecessary to be aggregated, and it can be evaluated independently and selected by
optimized selection model, that is, only one or a few feedback can win out. Competitive scientific
crowdsourcing has typical non-emergent property—volunteers complete tasks independently
and upload result, and then scientists or research teams select the most representative or optimal
feedback at last. Emergent property means that each individual contribution is a part of the total,
and then each will be aggregated and integrated from the sample quantity and distribution. Instead
of evaluating and optimizing individual contribution separately, more attention is paid to the
overall situation when individual feedback accumulates to a great volume.
In this regard, this paper divides emergent scientific research into quantitative emergence and
qualitative emergence from the perspective of value density. Quantitative emergence places more
emphasis on processing the arithmetic mean and statistical summary as to feedback samples
collected, and voting scientific crowdsourcing is a typical example. The assigner will not evaluate
or analyze any single vote result, but needs to conduct descriptive statistics and distribution test for