Training for mediators is a complex issue. It is generally effective for trainees to reflect on their past thinking, speaking, and acting. We present a text processing method which aids mediation trainees in reflecting on how they reached an agreement from their dialogue. The method is an improved variant of the Data Crystallization algorithm, which visualizes the inter-topic associations which foreshadow the intentional or unintentional subsequent development of topics far apart in time. We demonstrate how the dialogues which differ in the agreement quality affects the topological characteristics of the associations.
展开▼