A Study on Creativity Support System for Interactive Collaborative Work


ABSTRACT

This thesis summarizes the fruits of my research on systems that support creativity of interactive collaboration works, which has been done since June 1992 till March 1995 at ATR communication systems research laboratories, and since April 1995 till April 1998 at ATR media integration \& communications research laboratories.
Rapid changes in the structure of industry and the intensification of global competition in recent years have resulted in a situation where improvements in productivity alone are not adequate to develop enterprises. Accordingly, knowldege creation has recently become an essential task of enterprises. Furthermore, due to the drastically increased complexity and scale of recent projects, there is strong demand for tools that support group creative activity as well as individual creative activity. In such trends, many research efforts to enhance human creativity and to support effective knowledge creation of groups by utilizing computer and communication technologies which have drastically developed in recent years. This research belongs to this research area.
Focusing on the potential creative aspect of an interactive collaboration work, this research aims to establish basic technologies to construct systems which effectively support human creative thinking process, in particular human divergent thinking process, of interactive collaboration works. Discussions and conversations done by not only natural languages but also non-verbal mode, e.g. music, are included in the interactive collaboration works that are delt with in this research.
Systems to support human creativity can be classified into the following three categories based on the supporting manner: the secretarial level systems that provide electrical tools corresponding to a black board, cards and so on, the framework paradigm level systems that support structuring pieces of information input by users, and the generative level systems provide seeds of new ideas. A system of any level can effectively contirbute to human creative thinking. However, from the perspective of utilizing the contemporary computer environment, in particular the huge resource of the stored data and information all over the world, it is expected to realize systems of the generative level. The systems in this level can be divided in systems that provide information based on semantic relationships of words, and systems that provide information based on structural similarities of information. Currently, the former systems are more practical.
Therefore, many systems based on semantic relationships of words have already been developed. However, such systems still have the following problems: 1) How to effectively extract pieces of information which not only keep relation with the subject and context of thinking but also involve unexpected relations from different viewpoints. 2) How to intervene to a people's collaborative work to effectively stimulate their creativity. Many systems of the secretarial level support group discussions. However, there are no systems of the generative level which deal with group activity. 3) How to deal with the conversation by a non-verbal mode. Interactive collaboration works are executed not only by natural languages but also some non-verbal modes. Such works can be found in artistic domain, in particular in a creative work of music. However, there have not been any preceding researches to support such works. This thesis provides solutions for these problems.

This thesis is organized as follows.
In chapter 1, I clearly position this research and its purpose by analyzing the current status of the creativity support technologies and pointing out its problems in detail.
From chapter 2 to chapter 5, I discuss creativity support technologies for conversation done by natural language.
In chapter 2, an information retrieval method which can extract information to stimulate human divergent thinking is discussed. In a brainstorming session, it is often observed that pieces of information that are provided by an outsider whose knowledge domain is different from the other participants of the session often have effects to make a breakthrough in their fixed ideas. Hence, I model the process of information retrieval of an outsider, and implement it by using an associative memory technology. Moreover, from experiments with the implemented prototype system, I show that the system has ability to extract pieces of information which not only relate to the topic of discussion but also involve some different relations introduced from the outside knowledge.
In chapter 3, by applying a visualizing technology of structure of thinking or conversation based on a statistic method, I propose a method to extract information that involves some overlooked relations which are potentially included in a person's thinking or in a group conversation. Furthermore, by combining this method with the information retrieval method proposed in chapter 2, I implemented four kinds of information retrieval method to extract pieces of information which have various degree of relation to the topic. By applying the methods to experimental brainstorming sessions, I evaluate how pieces of information having different relations influence human divergent thinking. Additionally, it is also discussed when such a piece of information should be provided into a conversation to effectively influence.
In chapter 4, I discuss a non-delay topic-transition-detection method. Topic-transition is an important factor to determine the timing to provide a piece of information into a conversation. A conversation for creative thinking is usually free-conversation, i.e. conversation without any specific structures and evident goals, and its topics expand over various domains. Therefore, I propose a segmentation method for such conversations, which utilizes surface data obtained from utterances and time transition data through the conversation, but does not utilize any knowledge depending on specific domains nor any knowledge on structure of discourse.
In chapter 5, by integrating the technologies introduced so far and preparing multiple agents each of which supports a specific task of creativity or communication, I show a supporting system for creative aspects of conversations involving not only divergent thinking process but also convergent thinking process. Additionally, I discuss the effects to apply a multi-agent architecture for implelemting such applications.
In chapter 6 and chapter 7, I discuss creativity support technologies for conversation in music play.
In chapter 6, as a very basic support to communicate by music, I discuss a method that allows people who are not familiar with musical theory nor techniques for manipulating musical instruments to rather easily express their ideas in music. Focusing on the emotional functions of each note in music, I propose a new mapping method of notes onto an interface of a musical instrument, where a specific function of note is always assigned onto a specific position of the interface.
In chapter 7, focusing on the similarity between verbal communications and music communications, I propose a supporting method for creativity in music conversations by integrating the framework to support creativity in verbal communications introduced from chapter 2 to chapter 5 and the supporting method of music communication described in chapter 6.
In chapter 8, I summarize the fruits obtained in this research, and discuss the application areas and the future works.