Research On Innovation Ability Of Communication Industry Based On Complex Networks

The innovation ability is a topic all industries are discussing at present. How to improve the enterprise's innovation ability has become key link to adjust economic structure, transform growth mode, and improve competitiveness of the country and district. Therefore, the macro progress way of the enterprise's capacity for independent innovation through scientific appraisal to lead the industrial cluster to continuously improve the capacity for independent innovation has become one of important problems before us to be dealt with. In addition, complex network is a new subject, which has strong applicative background in different fields. Also for enterprise innovation ability of the research provides a new perspective.Based on the above problems, the article is to combine the innovation ability of enterprises in communication industry with complex network model, statistical property and detect community, which has certain studying value:This article is mainly work on the following aspects:1. Introduced the enterprise innovation capability of the research status and research enterprise innovation ability of meaning and the basic theory of complex networks makes a detailed narration.2. According to statistics theories, to apply principal component analysis and likeness coefficient theory to communication enterprises of our country to model complex network. This is to take communication enterprises of our country as samples, sort the data according to the category of questions in questionnaire, and reduce dimensions to overall property of the samples in principal component analysis method, and to draw complex network model with ucinet software after the similarity matrix between samples is obtained through data after reduce of dimensions, and analyze statistical property in the model with complex network theory so as to get useful conclusion.3. Aiming at special data structure in complex network model, to get number of lines between two nodes with depth-first search algorithm, and transform value of lines between two nodes to similarity between two nodes through self-defined likeness coefficient, and to get community structure of nodes through hierarchical clustering method.The innovation of this article is:(1) The hierarchical clustering method in traditional statistics is aiming at cluster sample and its indicator category. But due to particularity of data structure in complex network, this article defines the distance of a node to get the likeness coefficient of each node, which overcomes the insufficient of clustering algorithms hard to be directly used in organization detect in complex network, and has important theoretical meaning.(2) Put forward a complex network model based on communication enterprises. The model can well describe the innovation ability of communication enterprises of our country.