On Using Social Context to Model Information Retrieval and Collaboration in Scientific Research Community - Community-Built Databases: Research and Development - page 138

6.3 Background: Collaboration and Social Models
for Scientific Information Access
In this section, we discuss the integration of social context in order to achieve
retrieval and collaboration tasks in a scientific research community. In particular,
we first review major information retrieval models within scientific information
networks that neglect social context and then highlight the benefits of considering
several social relevance factors such as trustworthiness of authors, citation sources
importance, strengths between authors, and taggers' interests in order to enhance
the retrieval rankings.
6.3.1 Analysis of Coauthorship and Co-citation Networks:
How Strongly Is Connected the Scientific Community?
With the increasing collaboration between scientists, many researches have
addressed the structure of the scientific community and its evolution patterns.
These studies evaluate the cooperation among research groups and predict reliable
collaboration between local and international research teams. In fact, collaboration
does not cover only individuals, but also concerns research groups, institutions, and
countries. As defined in [ 34 ], the scientific collaboration involves two or more
scientists working on a research project and sharing intellectual, economic, and
physical resources. Having these various factors and the complexity of the interac-
tions during the collaboration process, it is difficult to represent and evaluate each
collaborator's contribution. Meanwhile it can be approximated through the coau-
thorship and citation associations explicitly defined in the resulting research docu-
ments. Scientific collaboration networks are extracted from bibliographic resources
and modeled using a graph where nodes represent authors or collaborators and
edges denote collaboration associations. In [ 35 , 36 ], the scientific collaboration is
represented by a coauthorship network where connections express direct and
collegial interaction between authors. By contrast, approaches described in [ 37 , 38 ]
focus on the citation network for modeling scientific collaboration as it represents
influence and knowledge transfer between authors. Notably, the analysis of the two
kinds of scientific networks shows an important collaboration within the researcher
community. It is concluded in [ 35 ] that authors of the experimental research fields
(biomedicine and astrophysics) tend to diversify their collaboration to reach 18
different collaborators. This study, covering several bibliographic databases, shows
also that the coauthorship highly connects scientists to include around 80-90% of
nodes in the giant component. Likewise, citation network analysis of physics
researchers [ 39 ] shows an important collaboration interaction with 74% of the
papers having ten or fewer citations and with an average of 14.6 citation links per
paper.