Conceptual approach to the development of technology evolution network based on structural and semantic analysis

Since technology plays a key role in new product/service development, many techniques have been developed to analyze technology evolution. Among many techniques, citation network analysis has been actively used for identifying technological trajectory. Even if there have been numerous studies to employ citation network to analyze technological evolution, it still remains a void in the literature how to differentiate each citation's role in technology evolution network. Generally, a paper or patent cites twenty to thirty references. However, most studies consider the importance of these citations as homogeneous, with less consideration on the importance of each individual citation. This can cause significant misleading for technological evolution analysis. Even if some studies tried to employ content-based citation analysis, these still remain conceptual. In response, this study suggests a new approach to analyzing technology evolution network by reflecting the role of each citation, considering both structural and semantic perspectives. For this purpose, we use natural language processing and machine learning technique to identify the importance of each citation by analyzing descriptive styles and semantic meanings. Thus, how each citation differs in terms of mediating technological knowledge flow can be measured, considering the frequency, structural position, and descriptive styles of each citation. Using different weight of each citation, a new way of network analysis is suggested to analyze technology evolution.