Modeling and Forecasting Numbers of Granted Patents by Using Published Applications

• Main Authors: Yi-Tung Chan, 詹益侗
• Other Authors: 陳達仁
• Format: Others
• Language: en_US
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/10146379298107910670

碩士 === 國立臺灣大學 === 機械工程學研究所 === 94 === Patent information is enormous and continuously expanding that makes it extremely useful in conducting technological forecasting. Published applications provide a preview of soon-to-come patents and earlier information exposure of new technology. The objective of this research is to find and build the relationship between the number of published applications and the number of granted patents. Furthermore, the results can be used in the forecast of the trend for specific technology in order to capture the future in that industry. Two short-term forecasting methods as well as one long-term forecasting method were developed based on the relationship between the number of granted patents and the number of published applications. These forecasting methods and the relationship between granted patents and published application were verified and established by the detail of three case studies. Comparing with traditional time-series ARIMA method, the predicting power of the short-term forecasting methods was similar. The long-term forecasting method modeled based on the characteristic of time lag between the patent granted date and its corresponding published application date was different from traditional long-term forecasting method and has shown superior predicting power.