Title: A Validity Analysis of Social Networks Extracted from Text Corpora: A Case Study on U.S. News Reports on China's Military Technology

Abstract: This paper aims to assess the representational validity of the social network extracted from a collection of texts. Taking U.S. news reports on China’s military technology as an example, this study compares the network generated from a collection of texts with a similar form of representation produced by human readers, and links the central nodes identified in the network to the key players in the historical events reported by the news articles. The result provides evidence of a moderate correlation between the matrix extracted from the corpus and the readers' representation of the texts. The nodes with high flow betweenness centrality in the networks correspond well with the key players identified in the news reports. The study suggests that corpus-based social network analysis can be a valid tool for analyzing social reality even though this new method needs further improvement.

Keywords: corpus, social network, validity analysis, news reports on military technology

Author: Xiaolei Ma, Professor, College of Arts and Sciences, National University of Defense Technology, Changsha, Hunan, China; Ping Li, College of Arts and Sciences, National University of Defense Technology, Changsha, Hunan, China.