Grounding in computer-supported collaborative problem solving
|
|  Yüklə 1.41 Mb.
|
The VCE: TecfaMOO
For our purposes of studying multi-modal computer-mediated collaboration, with an eye toward building a collaborative software agent, we needed a groupware system with at least the following components:
Our ideal system would integrate all of these facilities, including speech and typed text, organized into reviewable conversations, a whiteboard with flexible placement and editing of objects, with a facility for combining graphical objects into more complex schemata, with real-time pointing, and full mixed initiative between collaborators: either agent can do any operations at any time. The ideal system would also allow the importation of external data, such as predrawn figures, text, and icons or "stamps", and would be easily extensible to allow the logging needed for experiments, and the computer access by software agents. We several system systems which appeared as goog candidates.
Even with roughly comparable media, we noticed vast differences in the way different software allows one to interact with these media. In some cases, one can clearly say one functionality is better than another; but, in most cases, there is a tradeoff: how to make the best decision depends on the task that the collaborators need to perfrom, and even the way in which they might choose to perform this task. One must also consider the entire system, as many functions can migrate from one medium to another, depending on the relative costs and opportunities that a tool provides.
While we examined quite a few groupware systems, none of them (at least no public domain or moderately priced systems) met our ideal. First of all, many groupware systems are geared more towards management goals rather than synchronous collaboration. Also, for those systems using synchronous communication, much effort is currently being placed on audio/video links, rather than more persistent forms of communication, such as whiteboard drawings. The system we ended up choosing was a MOO environment. The has one advantage in that it is widely used on the Internet for collaborative, educational, and social purposes, and thus is fairly stable, and has a good base of ready users. Its most basic functionality is a text-based messaging system, providing several modes of communication. The MOO is a text-based a "virtual reality", including spatial locations and manipulable objects. The subjects move in different rooms, where they find suspects and objects. They talk to each other via two commands: "say..." to communicate with anybody in the same room, and "page John..." to communicate with John where ever he is. Importantly, for our purposes, it is also extensively programmable, so we could embed a problem environment, as well as special resources. MOO programming also allowed us to produce the protocols of actions and communications of the collaborations. The MOO works by sending information to a central server whenever someone hits There exist several hundreds of MOO environements on Internet. These experiments have been run with standard MOO called TECFAMOO, developed by Daniel Schneider and several colleagues at TECFA10. In this experiment the subjects use a MOO client called Mudweller which runs on Macintosh and TKMOO-lite a client which runs on UNIX workstations. The window is split into panes: a pane of 14 X 19 cm, which display about 60 lines of text (any interaction uses several lines) and, just below, a text entry pane just which enter to type 3 lines. Example 1 shows a part of MOO window. join sherlock Auberge du Bout de Nappe: Lower Corridor Obvious Exits: Lobby (to Lobby), UC (to Upper Corridor), B (to Bar), P (to Private Residence), R1 (to 1), R2 (to 2), R3 (to 3), and R4 (to 4). Auberge Guest Room: 1 You see Rolf Loretan and Claire Loretan here. Sherlock is here. Obvious Exits: Out (to Lower Corridor). Sherlock asks Claire Loretan about last night Claire Loretan answers "I was in the restaurant with my husband and the Vesuvios. When the restaurant closed, I briefly went to my room and then joined the others in the bar." Sherlock asks "Do you know when the bar has closed?" wisper Did you notice that he is an insurance agent? I don't understand that. "what are doing? You ask, "what are doing?" ask rolf about the gun hercule asks Rolf Loretan about the gun Rolf Loretan answers "i looks like a military issue gun. Why don't you ask that Colonel?" Sherlock says "Forget it. I thought it could help if we make a tab with the informations about where were th people at what time." "Actually sounds a good idea. You say, "Actually sounds a good idea. " "I think we should find more information about the gun You say, "I think we should find more information about the gun" Example 1: An excerpt from the MOO window. Bold lines are entered by the user (Hercule). We implemented several functions specifically for these experiments:
read heidi from dn2 You consult Detective Notebook 2. You turn Detective Notebook 2 to the page on Heidi Zeller. ---------------------------------------------------------- Info about Heidi Zeller : ---------------------------------------------------------- When asked about `last night' the answer was: I had a drink with Hans in the bar. Then, around 7.45, I went out with Lucie, first for a pizza and then to the new night club El Gringo. We were out pretty late because she was flirting with some Czech hockey players. When asked about `Mona Lisa Vesuvio' the answer was: What do you want me to say. She was one of those snobish businesswomen who believes that everbody is at her service. When asked about `gun' the answer was: I saw it when I was cleaning the Colonel's room the other day. He's a scary old man! Example 2: Retrieving information from the notebook (excerpt from the MOO window) |