Towards expanding the scope of trials via machine learning

Data analysis of issues on custody

Abstract

We investigate the possibility of expanding a trial process to the cases that are not usually considered in a court. Our central question is whether it is possible to extract points of argument on each case from the public opinions. To this end, for a few hypothetical cases, we conduct a free-response survey using a web application. The data is collected as a graph (network) so that the public opinions can be summarized using a graph clustering algorithm. As a result, we found that extracted points of argument sensitively reflect details of the cases.