Team Control Number
28747
2014 Mathematical Contest in Modeling (MCM) Summary Sheet
Methods of Measuring Influence Using Network Model
Summary
In this paper, we build the network model to measure the impact of researchers,
papers and so on.
We first use the network model to evaluate the impact of researchers,considering
researchers as the nodes, and using sides to describe collaboration among researchers.
We then propose the concept of importance degree and influence degree. They are all
the properties of nodes. Importance degree depicts a node’s own weight in the
network, while influence degree estimates a node’s total influence that mutual impact
among nodes is included. Based on the comprehensive consideration of the clustering
coefficient and degree, we put forward a new idea to measure a node’s importance
degree. Then combining with PageRank algorithm, we can evaluate the influence
degree of every node in the network. We can find ALON, NOGA M is the most
influential researcher. By analyzing the properties of nodes, we find that for a general
researcher, she/he should extend its collaborative network as greatly as possible,
especially partner researchers with high importance degree.
Second, imitating previous methodology, we build the model to evaluate the
impact of papers. We conclude that factors determining a paper’s influence contain
three indexes: the first author’s H-value, the journal’s Impact Factor and cited index
which is the concept we define to depict the influence degree of a paper in the aspect
of citation. Referring to the previous idea, we construct a new network reflecting
citation relationship among papers, then we can obtain the cited index of every paper.
After having collected the value of another two indexes, we use AHP to determine the
weight of three factors, and finally evaluate the total influence of a paper successfully.
We find that the paper Statistical mechanics of complex networks is most influential.
After that we view the other fields instead of the academic area to extend our
model. We apply our network model measure Chinese movie stars’ influence. We
select a greatly influential movie star to replace the status of Erdös, and construct the
cooperating network among movie stars. Having obtained influence degree of stars,
the ranking of influence of movie stars maintained a highly consistent with reality.
This is proof that our model is feasible. Then we discuss our model’s application for
academic, military and SNS fields roughly.
Finally, we make a sensitivity analysis for our model, and discuss the impact of
the changing of nodes and the papers’ feedback of citation relationship on the results.
Through previous analysis, we can see that our model can be applied to many
files, so it has a relatively high generalization.