共同出願特許ネットワークにおける高クラスタリング係数再現モデルの検証

概要

論文の詳細を見る
This paper investigates a model that generates networks with clustering coefficient as high as those in networks based on joint patent applications. A model has been proposed that generates a network structure that is similar to those of joint patent applications. However, the model is not sufficient to achieve a high clustering coefficient. The clustering coefficient is an index of network structures that shows the existence of groups that are densely linked. Our discoveries are summarized here as follows.(1) To improve the existing model, replacements of links are effective based on the exponential probability in degrees.(2) In the improvement, the degree distribution of the generated network has a different figure from the original network. In the above investigation, an original joint application network was necessary, and it was created from about five million Japanese patents issued between 1993 and 2002.