Knowledge Graphs – The Missing Piece in a Trial Team’s Arsenal

Any trial consultant or lawyer knows that every juror is different. We all have various experiences and interactions that shape our view of the world, and a common challenge in legal cases is trying to dissect a juror’s views and attitude towards the case at hand. It isn’t possible to have a long interview with each person in a jury pool, and even more difficult to gauge how jurors will interact with each other when trying to reach a verdict. However, in today’s digital-oriented world, we don’t have to. The vast majority of Americans have some sort of digital footprint, an online presence that reveals details about our lives and our views that consultants can use just so long as they can piece disparate bits of information together. 

When I took my first class on algorithms and data structures in college, the lecture that always stuck out to me was the topic of graph structures, or network graphs. Consisting of nodes and edges, network graphs can be used to represent any entity that can be connected to something else. Citation networks, social networks, biological networks, you get the idea. Even more fascinating as I moved into my graduate degree was the idea of a knowledge graph. A type of network graph, knowledge graphs consist of nodes (think people, jobs, hobbies, etc.) connected by edges that have a particular type of relationship (think Person A likes Person B, Person B voted for Person C, Person C works at Workplace A). Working quite similarly to the way we intuitively think, these structured representations of jurors can be a powerful tool not only to visualize an individual and key elements of their identity, but also to create networks of entire juror pools, social groups, ideological groups, etc.). 

For trial consultants, representing information in this way is not only useful for understanding individual jurors, but can also be used to understand how jurors will interact with each other. A combined network graph can immediately reveal that both Juror 1 and Juror 4 both follow some of the same people who work at a local brewery, or that Jurors 2 and 5 have both had rejected auto insurance claims. These connections are essential in gauging how jurors will interact, how they will group together or separately, whether they will align with each other or have antipathy towards others. 

Everything mentioned above doesn’t have to be restricted to solely jurors. Private investigators and trial teams, examining plaintiffs, defendants, or even non-legal cases, can use these tools to understand organizations, groups, or an important person of interest. The knowledge graph structure, which has been used to analyze everything from music lyrics to international diplomacy, is versatile and adaptable to any use case involving interactions of any kind. 

A couple of years ago, these sorts of graphs were painstaking to make, often taking hours to manually add in connections and run various algorithms. Today these graphs can be made automatically, both from collecting social interactions to generating nodes and relationships from normal human text. Not only can they be easily made, but they can also be easily combined with other sources. Say you have a network of where the jurors work. Using public data sources and their APIs, you can enhance such a network with information such as the average salary for each role, as well as past networks linking occupation, income, and political leaning to views on various societal issues that may be relevant to the case. 

These results can present an advantage to a trial team before even entering a courtroom, from understand the jury pool to having a better idea of which voir dire questions to ask. In a field where understanding human behavior can make all the difference, knowledge graphs offer a promising pathway to more informed, strategic jury selection.