Automatic Construction of Technology-Function Matrix for Patent Management: A Text Mining Approach

• Main Authors: Ming-Ruei Yang, 楊明叡
• Other Authors: Chin-Sheng Yang
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/31920813430851736217

碩士 === 元智大學 === 資訊管理學系 === 98 === Due to the rapid growth of knowledge economy, the value of intangible intellectual property becomes more and more important comparing with the tangible assets. Intellectual property has become the best competitive weapon in the commercial battlefield. Among the various types of intellectual property, patent is the most critical and important one. Appropriate patent management can benefit organizations greatly. There are various tasks involved in patent management. One of the most well-known ways for patent management is patent map. A patent map is a visualized expression of inventions in a specific technological field which offers tremendous business value by providing detailed knowledge of the scope of inventions within a target market and the necessary vantage for unlocking the true potential and strategic direction of your intellectual property. However, as the number of patents increases rapidly, it is not easy to conduct effective patent map generation which is labor-intensive and time-consuming. Consequently, many studies have concentrated on proposing effective method for patent map generation by incorporating emerging information technologies, such as data mining. The data-mining-based approach for patent map construction generally adopts the clustering analysis technique that is based on the analysis of the textual content similarities between patents and thus can only generate a one-for-all expression for potential users. However, different users may have diverse preferences on the expression of patent map. In response to the limitation of existing techniques which are incapable of providing tailored patent map, we employ the text mining technique and propose a method for patent map generation which considers users’ preferences. Since the generated patent map is generally organized and represented as a two-dimensional matrix, we call the map technology function matrix. According to the results of our empirical evaluation, the proposed text-mining-based approach can easily generate the desired technology function matrix. Moreover, considering the difficulty of technology function matrix generation, the effectiveness of the proposed technique is promising.