Scientific and industrial keyword analysis using structured covariance and clustering

Using scientific and industrial keyword analysis (SIKA), many industrial companies have built their research and development (R&D) strategies for improving technological competitiveness in market. The technological keywords extracted from journal papers and patent documents are good resources for SIKA. In this paper, we use patent keyword data as scientific and industrial keywords for SIKA. A patent contains various information of developed technology such as patent title, abstract, date, citation, etc. Because the exclusive rights of technologies applied and registered to patent system are protected by patent law for a certain period. We also consider statistical methods for the SIKA. First we perform technology clustering using K-means clustering of technological patent keywords. Next we carry out the principal component analysis (PCA) from the clustering results. Using the first and second principal components, we obtain PCA plots for techno-logical clusters. So we can understand the technological structure of given and target technology from the PCA plot results. Combing the technology clustering and PCA plots, we propose a method of SIKA to build valuable R&D strategy of company. To illustrate how the proposed method could be applied to real problem, we make experiments using many technological keywords for given technology field.