A New Method of Creating Patent Technology-Effect Matrix Based on Semantic Role Labeling

Patent technology-effect matrix is very helpful for patent analysis but difficult to create. In addition to manual methods, most existing methods either use IPC and statistical classification or perform pattern matching algorithm to extract technology phrases and effect phrases. They are easy to ignore implicit semantic information and cannot mine deeply intelligence in patents. This paper retrieves patents in related domain, collect the patent texts, and apply semantic role labeling to create technology-effect matrix. Experiments show that semantic role labeling achieves satisfactory results and save manpower.