Database Reference
In-Depth Information
On the basis of the coauthorship network, a set of indicators are introduced by
bibliometrics to quantify scientific collaborations such as the coauthorship index
that represents the average number of coauthors per paper and the collaboration rate
that calculates the rate of papers resulting from collaboration between two groups of
scientists. In most of these measures generally rank institutions and research
groups, it is proposed that authors to be ranked according to their scientific quality
by applying the PageRank algorithm on the coauthorship network [ 40 ]. We note
that coauthorship is considered either as a symmetric association that equally
involves the collaborators [ 41 ] or as an asymmetric association that depends on
the frequency and the exclusivity of the coauthorship between each couple of
authors [ 40 ].
Scientific collaboration is also quantified based on the citation links that repre-
sent indicators of scientific excellence. According to this purpose, the H-index [ 42 ]
measures the scientific quality of an author based on his most cited papers and the
number of citations that he has received. The approach presented in [ 38 ] proposes to
weight the citation link depending on the coauthor numbers, and then rank authors
by disseminating scientific credits on the network through the citation links to
simulate knowledge redistribution between scientists. However, the citation feature
is not sufficient to estimate paper relevance considering that the probability of
jumping to an article is proportional to its publication time. The citation count is
extended so as to rank papers according to their age and expected citation over time
later [ 43 , 44 ]. Accordingly, CiteRank [ 45 ] takes into account publication time to
rank the articles and defines a random walk to predict the number of expected
citations.
6.3.2 On Applying Social Networks for Literature Retrieval
and Scientific Collaboration
An information retrieval within bibliographic resources differs from other usage
context by a specific information need targeting high scientific quality of retrieved
documents in addition to the query similarity. Therefore, scientific indexes and
academic digital libraries, from the introduction of the S HEPARD'S C ITATIONS (1873)
to the recent of G OOGLE S CHOLAR 2 (2004), have addressed one common issue:
evaluating the scientific quality of bibliographic resources. To tackle this problem,
literature access ranks documents using bibliometric measures such as the citation
index. Other approaches apply hyperlink analysis algorithms to the citation network
of documents and rank resources depending on their authority computed by the
HITS and PageRank algorithms [ 46 , 47 ].
In fact, the quality of a document is related to its author once document and
authors are inseparable entities and represent each other. On the basis of this idea,
2 http://scholar.google.com/
Search WWH ::




Custom Search