Clustering method utilizing various kinds of relationship and locality

摘要

In order to get useful information from plenty of documents existing as digital data, to distinguish and visualize this information is necessary technology. Traditional keyword clustering is superior to other clustering in objectivity for data, however, cannot deal with data locally and subjectively. Other clustering as KJ method, on the contrary, implementing these problems cannot cluster plenty of data. In my paper, we suggest various kinds of clustering methods to fulfill more intimate objective clustering. In order to implement locality, we discuss it in the light of deciding keyword and partly clustering, and design a system fulfilling these methods. We suggest methods about fulfilling subjectivity.