Enhancing technology clustering through heuristics by using patent counts

International Patent Classification (IPC) system is a hierarchical classification structure used essentially to classify and explore patents along with the technical fields which they are concerned with. Therefore, corresponding number of patents for a certain IPC, can serve as an indicator of technical developments in the relevant area. These numbers can also form a basis for investigating state of the art for a particular field of technology. This paper proposes an approach for clustering of patents for those of the technologies listed by IPC via the number of patent counts. A set of n real numbers indicating the patent counts for different technologies is partitioned into k clusters such that the sum of the squared deviations from the mean-value within each cluster is minimized. With this purpose in mind, two different heuristics have been considered for clustering since complete enumeration would take considerable solution time. The first heuristic is specifically proposed for this study and the second one is Great Deluge Algorithm (GDA) which has been extensively used for solving complicated problems. The proposed heuristics are coded in visual basic (VB) 6.0 and a user interface is developed for the program. The developed program attempts to find the appropriate k value in order to make the best possible clustering. As an application of the proposed clustering approach, patent data that is retrieved from web site of Turkish Patent Institute (TPI) has been used for clustering technologies.