Congressional Vote Analysis Using Signed Networks
2018 18TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING WORKSHOPS (ICDMW)(2018)
摘要
In today's era of big data, much can be represented as a network. However, most of the work in traditional network analysis is unable to handle many existing network types, which is due to certain networks having added complexities. For example, signed networks, which have both positive and negative links, have been shown to require dedicated efforts due to the methods designed for typical unsigned networks (those having only positive links) being no longer applicable. One specific type of signed network is that of voting records, such as the Senate and House of Representatives from the U.S. Congress, which form signed bipartite networks between the congresspeople and the bills voted upon. With the current tensions between the two prominent political parties in the U.S., it seems time to ask the question if signed network analysis methods are able to aid in our understanding of the underlying dynamics of the voting habits in the U.S. Congress, since they drive some of the most influential decision making processes in the country. To this end, in this paper, we conduct a thorough analysis on the behaviors of both current and past U.S. Congress voting datasets uncovering numerous patterns, extending and then investigating the applicability of balance theory in the signed bipartite setting, and then finally leverage our findings to accurately predict the sign of missing links.
更多查看译文
关键词
Signed Networks, Balance Theory, Signed Bipartite Networks, US Congress
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络