Monitoring the key technology trends by combining chance discovery and survival analysis: Study on solar cell patent documents

Patents provide objective and rich information on technology, so it becomes a valuable source to monitor. The fact that the technological information in patents is almost stored in text format makes the monitoring task a hard work. The purpose of this study is to introduce a new approach which combines chance discovery, one of a text mining techniques, and survival analysis to explore the emergence of technology terms and to identify their potential. This study applies the chance discovery to extract technological terms which are important but occur infrequently. Moreover, this study conducts longitudinal analysis by employing survival analysis on the occurrence time for the extracted technological terms. The hazard rate is used as an indicator to identify the potential of the terms. This study collected solar cell patent documents from both of issued and applied databases to perform the proposed approach. The patents of 2001 to January 2005 serves as experimental sample to calculate the hazard rates of extracted technological terms and those from July 2005 to 2008 are for validity testing. The results reveal that there is a high correlation between the hazard rate and the patent numbers for extracted technological terms.