Abstract
In this thesis, we proposes the technique for analyzing the process of change that occurs in a data stream. The process of a change refers to topics generated in data and indicates the development of data flow after these topics are generated. The present study assumes the data stream to be a data set where data with graph geometry changes timewise; further, this study analyzes the graph system that shows the process of the temporal variation of the data stream. In particular, the change in the extracted community of each graph of an affiliated inside is analyzed with consideration of the volume of data. The technique proposed in our previous study has considerably influenced the extraction of the process of change for the time interval in which each graph with an affiliated inside was made. Further, we propose a method for analyzing the process that uses the interval time parts can be obtained by designing the kernel function that considers the time change of the community, and the process of changing the longer time frame can be extracted.
