Helper Character for a Virtual Meeting-Place

Providing online meeting places with video and motion support does not guarantee the successful formation of community, or even of conversation. People need to find things to talk about, and get over initial awkwardness, to really be comfortable communicating in such meeting spaces. The helper character project goal is to create an character that can assist people in forming groups and beginning conversations. We believe this will improve the experience people have in the virtual meeting place. Also, we believe the character may be especially helpful in situations such as cross-cultural communication, where visitors have even less common context to rely on in beginning their conversations.

Here is the Helper character. Click on this image to view a sample Quicktime animation of the character. We implemented 6 animated poses for the character for our prototype. To see Quicktime animations of all 6 poses, click here.

Here is a screen-shot of the first draft of our system in use. The agent uses a text window to ask questions and offer suggestions. Participants can see and hear each other, and can move freely in the virtual environment using their mouse or arrow keys.

There are several features we plan to implement in this character. Currently, we have the first feature implemented, and are conducting a cross-cultural study between Kyoto University and Stanford University with our prototype agent:

• Get people talking by offering safe topics:
  If people are silent, the character will ask a short series of questions of the participants, so as to uncover a common topic for them to talk about. Topics are selected from a list of known safe topics for people from the U.S. and Japan, since we are testing out how well the character can assist cross-cultural communication between these two groups.
• Help create groups:
  If the character notices more than one person alone in the system (not in a group) it will try to get them to group together.
  
• Help people join groups:
  If the character notices one person standing outside a group, it will ask if they might want to join a group, and if so, which one. The character asks questions of the person to get a topic going, and then brings these answers to the people in the group, to ask if they want the person to join. If a majority do, the character brings the person to the group.

Toward creating these features, we have already built awareness of silence into our character, and are building awareness of groups in as well. We have implemented the ability to ask questions, using safe topics that we collected from Japanese and American university students.

Cross-cultural investigation
We are currently investigating whether the character's help in introducing cross-culturally safe topics of discussion helps people from different cultures to feel more comfortable and have better and less stereotype-driven or awkward interactions. We are running an experiment between NTT and Stanford University, using a broad-band dedicated connection that allows us to support high-quality video display. This experiment also has an unsafe topic condition, in which the agent offers awkward topics to the conversation pairs. We will finish running our study at the end of May, and hope to report results of this experiment by October.

team members: Nakanishi and Isbister, with the help of Kyoto University Department of Social Informatics students, as well as Stanford University Department of Communication Professor Clifford Nass, Ph.D. student Eva Jettmar, and other students at Stanford within that department.