Patent Citation Network Simplification and Similarity Evaluation Based on Technological Inheritance

In recent years, the speed of technology development and the number of issued patents are increasing rapidly. The relationships among patents, especially citation relationship, are getting more and more complicated. Hence, how to utilize patents and their citations effectively for studying technology development and assisting innovation is a significant research topic. In this article, we propose a framework to simplify complex patent citation networks and evaluate the similarity among patents based on technological origins and claim texts. First, we adopt the term frequency-inverse document frequency technique to convert patent claims into vectors. Second, based on these vectors, we compute the claim similarity between two patents with a citation relationship as a measure of technological inheritance. Third, the citation networks are simplified from the perspective of whole and maximum technological inheritance, respectively. Fourth, we take indirect citations and claim vectors into consideration for evaluating the similarity of any two patents on the aspect of technological elements. Finally, our framework is applied to surgical robot domain to reveal the development trend of technological inheritance and evaluate the technological similarity among patents. The main scientific contributions of this article include the following: First, we simplify patent citation networks from the perspective of technology based on patents' claims and their citations, which can be adopted in important patent detection and patent clustering, etc; Second, the similarity of two patents is evaluated in a more detailed way by considering their claims and the second order citing patents.