Structural Analysis of Patent Claims for Assessing Technological Change

This study proposes a method for analyzing technological change based on the internal structural characteristics of patent claims. Focusing on the hierarchical relationship between main and dependent claims, the method examines how technical elements are organized within claims over time. Compound words are extracted as representations of technical elements, and four structural indicators are defined: total number of claims, number of main claims, lexical overlap between main and dependent claims, and similarity among dependent claims. Using these indicators, patent claims are classified into structural types through standardization and hierarchical clustering. The proposed method is applied to U.S. patent data in the B60K automotive technology field. The results show that, although average differences in individual indicators are limited, the distribution of claim structure types changes significantly over time. Compared with conventional patent analyses based on citation relationships or patent counts, this approach captures technological change through differences in claim structure, providing a structure-based analytical framework for understanding technological change.