1.1. Social Networks and Assumptions about Place and Time
A social network describes a group of people who share some sort of social connections, whether through work, or friendship, or otherwise. The social network concept stems from mid 20th century sociology. Alignment of social network studies with computer science in about the 1970s allowed the connections among individuals to be weighted and computed mathematically on a large scale, with weights indicating, for example, strength of the relationship.
An analysis of a social network generally focuses on the groupings of people. The people might be employees of an organization, for example, or colleagues in a discipline, sportsmen on teams, or characters in a novel. Questions that could be answered by the analysis include: Which individuals are in what group? Who leads each group? Who is second to the leader? Who is between two groups? Are group relations friendly or antagonistic? At the foundation of the social network literature are Scott (1991 [2004]), and Wasserman and Faust (2004).
Social networks have been diagrammed with people as points and their social relations as lines. The points are called nodes; the relations are called ties or edges. The groups are cliques. The aggregate of cliques form a network at some time. Most social network studies and the standard social network diagrams account for neither space nor time. This may be because it is assumed that the network is spatially and temporally persistent and so does not change, or because a network snapshot is good enough. Whichever assumption is made, often space and time are treated as irrelevant factors.
Our social networks in our research are formed from connections among people mentioned in news articles. The activities of people and their co-relationships in space and time come from the context of the news story. When we extract names of peoples from these texts and build links among them using the proximity of their names in the article, we are in essence attributing a relationship among the people. This context can be characterized by the spatio-temporal setting of the news article.
Social network analysis of vast amounts of text through data mining, as described in this research, affords an overview of events. No reading of the text is necessary. This does not simply save an analyst much time and effort; it allows assimilation of text on a scale that would otherwise require many people to analyze. Even though errors occur because network nodes are only inferred, automated node extraction saves times and offers insight that is valuable. We suggest that the accuracy of the extracted network can be enhanced through improved extraction of spatio-temporal information that describes the network membership and relations among members.