![](https://d3ilqtpdwi981i.cloudfront.net/VgKKU_BqnUwHuvqPSeeRr3sf-wY=/425x550/smart/https://bepress-attached-resources.s3.amazonaws.com/uploads/70/a4/9f/70a49f96-7838-4355-b844-beb79101768b/thumbnail_5e60c09e-a0d3-47b1-a0d3-cc9526215a3b.jpg)
The construction of knowledge networks from text is a novel way to study the cognitive organization of domain specific content. This dissertation evaluated the application of knowledge networks to legal text. Text analysis methods were used to transform text from 8,014 Supreme Court opinions into matrix data suitable for the construction of knowledge networks known as SCOD networks (Supreme Court Opinion Derived networks). Four specific hypotheses were then tested to better understand the meaningfulness and validity of SCOD networks. The first hypothesis considered differences between SCOD networks and random networks. The remaining hypotheses considered the ability of SCOD networks to reflect known issues of the Court. Monte Carlo simulations, various graph theoretic measures and measures of graph similarity were used to test these hypotheses. Results showed significant structural differences between SCOD networks and random networks. SCOD networks were also shown to have good face validity in representing scholarly characterizations of the Supreme Court, and in particular reflected known issues concerning the influence of ideology on Supreme Court decision making. In general, this work demonstrates the potential in using knowledge networks to help answer a wide variety of questions concerning Supreme Court decision making.
- knowledge representation,
- text analysis,
- supreme court
Available at: http://works.bepress.com/anne-lippert/1/