Reflective visualization of dispute resolution Yoshiharu Maeno Social Design Group Bunkyo-ku, Tokyo 112-0011, Japan. maeno.yoshiharu@socialdesigngroup.com Katsumi Nitta Tokyo Institute of Technology Yokohama-shi, Kanagawa 226-8502, Japan. nitta@ntt.dis.titech.ac.jp Yukio Ohsawa Tokyo University Bunkyo-ku, Tokyo 113-8563, Japan. ohsawa@sys.t.u-tokyo.ac.jp Abstract—Mediation is a form of alternative dispute resolution which aims at assisting disputants in reaching an agreement on a disputed matter. Debate ensues on what skills an individual needs to play a mediator’s role effectively. Education and training for mediators become complex issues. Then how can mediator’s skills be trained in spite that the skills can not be defined clearly? In cognitive science visualization of and reflection on one’s behavior is proven effective in such a situation. In this paper we explore a text processing method for reflective visualization of a dialogue. The dialogue is a time sequence of utterances from a mediator and disputants. The method visualizes an inter-topic association which foreshadows the intentional or unintentional subsequent development of topics indicated by temporal topics clusters far apart in time. The method is applied to a mediation case where a dispute between a seller and a buyer on cancelling an purchase transaction at an online auction site is resolved. Index Terms—Dispute resolution, Mediation, Reflection, Visu- alization I. I NTRODUCTION Resolving a conflict between parties having opposing opin- ions is an important social requirement. Mediation is a form of alternative dispute resolution, which refers to a rather private and confidential extrajudicial process. Mediation aims at assisting disputants in reaching an agreement on a disputed matter. Companies often hire mediators in an attempt to resolve a dispute with workers’ unions. Mediation is different from arbitration where an arbitrator imposes a solution on the disputants. Rather, a mediator uses appropriate skills to improve the dialogue between the disputants and find solution. Mediator’s skills range widely from the ability to remain neutral, the ability to move the disputants from the impasse points, to the ability to evaluate the strength and weakness of the disputants correctly. Therefore, education and training for mediators become a complex issue. Debate ensues on what personal attributes an individual needs to play a mediator’s role effectively. The necessary skills and personal attributes have not been identified clearly yet. However, appropriate means is necessary to education and training for mediators, as resolving a dispute by mediation experiences increasing acceptance and utilization. The idea of reflection is a clue when we need to improve a skill which can not be defined clearly and taught by trainers. Reflection in cognitive science [12] and computer- mediated communication [15] means the ability to recognize and understand oneself, discover something unexpected, and create something new [6], [13], [14]. Particularly, visualization of the past utterances, decision-makings, and actions is one of the most practical tools to foster reflection. For example, presenting the design work until obtaining the intermediate outcome of an art piece in the form of a graphical diagram is proven effective in improving the quality of the final output in a university education program of creativity [5]. Reflective visualization and verbalization are proven effective in helping a person become aware of his or her unconscious preferences [10]. We expect that such reflective visualization is also promising in education and training for mediation trainees. Utterances are relevant and convenient information records for the trainees to reflect on. They are essential inputs to negotiation log analysis [11] and online agent based negotiation assistant system [16], [17]. Similarly, mediators and disputants can reflect on themselves by looking back the way how the dispute was resolved in a dialogue. In this paper, we explore a text processing method of the dialogue for reflective visualization and apply it to a mediation case. This paper is organized as follows. The text processing method is proposed in II. After defining a dialogue as a time sequence of utterances from a mediator and disputants, it describes the method to derive temporal topic clusters and inter-topic association from the recorded utterance texts, and draw a graph-structured diagram which makes the clusters and associations visible clearly. The inter-topic association foreshadows the intentional or unintentional subsequent devel- opment of topics indicated by temporal topics clusters far apart in time. Demonstration of the method using a mediation case is presented in III. The method visualizes the way how mediation trainees resolved a dispute between a seller and a buyer on cancelling an purchase transaction at an online auction site. The resulting implications and future works are discussed in IV. II. METHOD A. Dialogue The dialogue d is a time sequence of the recorded utterance texts u t from a mediator and disputants. It is represented by eq.(1) formally. The subscript t means the time when the utterance is observed. We do not use the absolute time from the beginning of mediation. Instead, the i-th utterance from the beginning is associated with an integer time t = i approximately. In eq.(1), T is the number of utterances in mediation. d =(u 0 ,...,u t ,...,u T −1 ). (1) Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics San Antonio, TX, USA - October 2009 978-1-4244-2794-9/09/$25.00 ©2009 IEEE 1760