Methods for Extracting Changes in Human Relationships on Web Based on Commonality Analysis of Graphs

Along with the rise of intellectual activities archived on the Internet, it has increasingly been important to understand human relationships on the Web. Understanding the connection between people is important in deeply exploring a particular person and information about the person. However, there is a limit to understanding human relationships manually due to Web bloat. Moreover, relationships greatly change with the passage of time, such as when the organization to which a person belongs changes. These factors have made it difficult to understand relationships. Against these problems, there have been several methods for visualizing human relationships as networks. However, they have not been effective enough because target people and connection factors are limited, and changes in relationships are not taken into account. Therefore, this research is aimed at developing novel support for understanding human relationships. In this paper, we propose methods for dynamically extracting changes in human relationship networks on the Web. Specifically, we initially develop methods for extracting relationships in accordance with the person of interest and relationship factors. We then develop methods for expressing the extracted human relationships as a network. Moreover, we develop methods for extracting the common parts of two human-relationship networks and significantly changed portions by comparing the networks. We conducted experiments and discuss the effectiveness of the proposed methods on the basis of the results.