摘要
Patent analysis depicts the innovative directions of research and development (R&D) and technical breakthroughs in given domains, because patents are effective means of intellectual property right (IPR) protection for innovations with viable commercial applications. Patent analysis often helps companies monitor IPs and assess their competitive positions in the marketplace. Patent clustering is one of the crucial analytical approaches in the overall patent analysis procedure for gathering the relevant technologies into coherent and specific clusters. Prediction methods for future technology directions are then utilized to formulate effective R&D strategies and to forecast likely outcomes. Currently available approaches in technology forecasting through clustering are mostly based on general patent text mining to identify key words/terms in the patents and group patents according to similar appearances of keywords. These approaches are not particularly effective at categorizing patent features. A company’s R&D or IP decision makers cannot fully rely on the categorization analysis outcomes to obtain precise insights into emerging technological trends. This paper proposes a new approach to patent claim-based technology clustering to forecast the frontiers of IP protected technologies. The proposed approach is applied for strategic patent analysis of Laser Interference Lithography (LIL) innovations. Patentable features are identified from annotated elements within independent claims of patents. These features are then summarized into simplified sentences for the definition of specific clusters. This method establishes a unique clustering principle to improve patent analysis accuracy and credibility of forecasting future R&D trends based on crucial patent claims.