Technology and Effect Matrix for Patent Clustering

Patents contain important technical, legal and economic information. Their annual publication accounts for a quarter of the books and journals in the world. Along with the growing number of patents, patent clustering analysis is becoming more and more important. We focus on two key issues in patent clustering, namely patent representation and data visualization. Firstly, patents are represented as technology and effect pairs, and then patents are clustered based on technology and effect matrix, finally a multi-layer patent map is generated. The experiments results show that our method has higher efficiency and better clustering effect than traditional vector space model and the visualization of clustering result is more practical and scalable.