Analyzing trending technological areas of patents

Abstract

Analyzing technological areas of inventions in patent domain is an important stage to discover relationships and trends for decision making. The International Patent Classification (IPC) is used for classifying the patents according to their technological areas. However, these classifications are quite inconsistent in various aspects because of the complexity and they may not be available for all areas of technology specially the emerging areas. This work introduces methods that applied on unstructured patents texts for detecting accurate technological areas to which the invention relates, and identifies semantically meaningful communities/topics for a large collection of patent documents. A hybrid text mining techniques with scalable analytics service that involves natural language processing which built on top of big-data architecture are used to extract the significant technical areas. Community detection approach is applied for efficiently identifying communities/topics by clustering the network graph of technological areas of inventions. A comparison to the standard LDA clustering is presented. Finally, regression analysis methods are applied in order to discover the interesting trends.