The Dynamic Analysis of Patent Tagging Networks

Yueh-Hsuan Weng, National Chiao Tung University

Abstract

Complexity science attempts to explain structures and rules governing systems which take organized but unpredictable form. Two important areas of research include network structure and dynamics. Regardless of whether the fundamental elements under examination are biological, social or physical in manner, they can often be understood in network form. Common examples include disease epidemics, electric power transmission, ecological food chains, academic citation networks, and of course the internet, all of which are complex networks somewhere between random and ordered on the complexity scale. Following up on recent network theory research, the author will systematically analyze the dynamics of complex legal networks using social network analysis with a special focus on patent documents. Patent citation networks are generated by unidirectional citations among patents. Analyzing these links typically requires computerized natural language processing (NLP). However, the use of NLP in analyzing patent networks has several limitations. The author has therefore adopted a new approach using the connecting relationships between technical codes and patents under the Multi-Dimensional Patent Classification System used by the Japan Patent Office (JPO). This system is used to construct patent tagging networks, which can then be subjected to three further analyses examining the fundamental characters of the tagging networks, the dynamics of patent unit networks and the dynamics of technical code unit networks. The author will first explain the construction of a patent tagging network before moving on to discuss applications, including dynamic network analysis.