Community Analysis and Maintenance Functions in the Public Opinion Channel
Tomohiro Fukuhara, Masaki Chikama, Toyoaki Nishida
The Internet enabled us to find information and people easily. We can find people who have similar interests or goals, and form or join a network community. A network community has possibilities for creating artifacts, sharing knowledge, and solving problems more effectively than traditional organizations such as companies and universities. Linux and free software communities are good examples of successful network communities. We consider that fine interactions, which enable community members to create and share knowledge effectively, exist in those successful communities. We call those communities knowledge-creating (KC) communities.
Besides KC communities, there are sick communities where too few or too much information is exchanged. There occurs a flaming, which is an endless quarrel between community members, frequently in a sick community. In another community, the deflation of communication called spiral of silence hinders exchange of opinions There are few fine interactions in sick communities
Our concerns are factors that form and sustain a KC community. What are differences between KC and sick communities? How can we form a KC community? What kind of interaction sustains a KC community? Although we don't have clear answers for these questions, we are tackling them by creating and evaluating a community support system called PUBLIC OPINION CHANNEL (POC).
Through development and experiments of the POC system, we found the importance of an experiment stage in the development cycle of a community support system. We can acquire valuable data in an experiment. Based on acquired data, we can improve the system and have a next experiment. Meanwhile, having an experiment is not easy because maintaining communities and analyzing data are burdens for researchers. Functions for supporting researchers to have an experiment smoothly are needed for investigate a KC community efficiently.
In this paper, we propose community analysis and maintenance functions of a community support system. We implemented those functions in the POC system. Through a field-test and a social psychological experiment using the POC system, we found feasibilities of the proposed functions for having an experiment smoothly.
2. Community analysis and maintenance functions in a community support system
In this section, we describe requirements for community analysis and maintenance functions by analyzing the development cycle of a community support system.
2.1 Development cycle of a community support system
To investigate a KC community, repeating a cycle that consists of implementation and evaluation of a community support system smoothly is important. Figure 1 shows the development cycle of a community support system.
The cycle consists of (1) analysis, (2) design, and (3) experiment stages.
(1) Analysis stage.
In this stage, an analyst observes activities in a community, and specifies requirements for supporting those activities. S/he tells the requirements to a system designer for designing a system.
(2) Design stage.
The system designer designs and implements a prototype system according to the requirements. The system is improved in case of an existing system. The system is tested in the experiment stage.
(3) Experiment stage.
The analyst cooperates with a community organizer to have an experiment. During an experiment, the analyst observes activities of community members, and records data such as utterances and operations of members. The organizer supports the analyst by doing miscellaneous works such as creating new communities, configuring settings of them, adds or removes accounts of community members, and removes unnecessary messages and communities. The analyst analyzes data, and gives feedback to the designer for improving the system.
Repeating this cycle smoothly is important for both of development of the system, and investigation into a KC community.
2.2 Problems in an experiment of a community support system
Having an experiment is not easy because analyzing data and maintaining communities are burdens for an analyst and a community organizer. In a long-term field-test that is held for several months, a large volume of data is acquired. To find important facts from those data is not easy for an analyst. Furthermore, it is difficult for an analyst to follow the latest state of a community during a field-test because s/he is busy observing activities and recording data. Furthermore, s/he might also have to maintain communities.
From the viewpoint of a community organizer, s/he has to engage in miscellaneous works during a field-test. An organizer has to prepare communities before a field-test. S/he also has to prepare community members’ accounts (for account-based systems). During a field-test, an organizer has to check messages posted by members. In case of inappropriate messages, an organizer has to remove them. Without an easy interface for configuring settings of a community, an organizer has to edit configuration files of the system manually. To edit configuration files, an organizer has to have much knowledge on the system such as internal structure for changing settings of a community. It is not practical for an organizer to have such miscellaneous knowledge because s/he is not a designer.
Community analysis and maintenance functions are needed for having an experiment smoothly. Previous works on field-tests of community support systems, however, do not mention those functions. Although frameworks of a community support system are proposed, their focuses are not on community analysis and maintenance works. The lack of community analysis and maintenance functions hinders system development and investigation into KC communities.
Community support systems should have community analysis and maintenance functions. Those functions should facilitate (1) an analyst to collect and analyze data, and (2) a community organizer to configure settings of a community.