A Clustering Method of Highly Dimensional Patent Data Using Bayesian Approach

Patent data have diversely technological information of any technology field. So, many companies have managed the patent data to build their R&D policy. Patent analysis is an approach to the patent management. Also, patent analysis is an important tool for technology forecasting. Patent clustering is one of the works for patent analysis. In this paper, we propose an efficient clustering method of patent documents. Generally, patent data are consisted of text document. The patent documents have a characteristic of highly dimensional structure. It is difficult to cluster the document data because of their dimensional problem. Therefore, we consider Bayesian approach to solve the problem of high dimensionality. Traditional clustering algorithms were based on similarity or distance measures, but Bayesian clustering used the probability distribution of the data. This idea of Bayesian clustering becomes a solution for the problem in this research. To verify the performance of this study, we will make experiments using retrieved patent documents from the United States Patent and Trademark Office.