| TECFA | Home Search News |
| Overview | People | Teaching | Projects | Resources |
It has been shown that metacognitive reasoning helps individuals in problem solving. Metacognition one one hand refers to the knowledge one has about his own mental processes and on the other hand to regulation and control processes of the action (Schoenfeld, 1987).
The distributed cognition approach (Salomon, 1993) considers a group of persons and the tools they use as a single cognitive system and Nickerson (1993) adresses the question if such a system also has metacognitive skills.
In a group setting, metacognition not only covers strategic reasoning related to the task (what to do next, evaluation of the work done so far, ...) but also reasoning related to the interaction itself.
Now the question is: are people aware of the way they regulate the collaboration and the task ? A priori, I don't think so, but I hypothesize that a better regulation of the interaction leads to improved performance in terms of a better problem-solving strategy.
This hypothesis might be checked by observing the effects of regulating the interaction between people on their problem-solving strategy. One way to regulate the interaction consists in having a human observing what is going on and telling people how to behave. Another approach is to use a 'computer agent' to supervise and intervene when necessary. A third approach, the one I chose, consists in triggering the regulation of the interaction by providing the users with reflective tools.
The task I study consists in solving a traffic-light problem, i.e. people have to tune the green-red periods in a way to minimize the car's waiting time at intersections. I chose this task because it is open-ended, i.e. there is not one correct solution. Sometimes, in previous experiments (where people had to set up a conference schedule) the solution was too obvious and one member of the pair was solving it alone.
In the experimental situation subjects will be given a tool which represents their performance as the mean time cars need to cross the city. In one of the conditions, the tool also shows indicators about the dialogue and/or the collaboration. These indicators are computed on the fly during the interaction. Dialogue indicators consist in acknowledgment rate and delay, count of utterances, type of utterance (possible if a structured dialogue interface is used),... Collaboration indicators reflect the partition of work between subjects, for instance, the relative proportion of problem-solving acts performed on the lights by each subject (distribution of work), a comparison of communicative acts vs. problem-solving acts for each subject (roles).
Concerning the methodology, I'm still looking for a formalism to represent interaction patterns. I've been looking at state-transitions nets which might be an interesting path to explore. All the possible actions-moves people can do are represented as states. The trace of activity corresponds to a path from one state to another through transitions. You then can compute probabilities associated to each transition and compare these probabilities across experimental conditions. For instance, the transition from 'implement' to 'discuss' might reflect if people are stuck in implementing a solution or if they (more or less often) take some distance and discuss what they've done.
|
|
The same in VRML (1) (2) (Thank you Dao for the VRML expertise ...)
I tried both and finally took JSDT because I did not need a very general solution and having a traditional client-server architecture was ok. Only needed 2 clients connect to a server.
I tried out JavaSim from the University of Newcastle but got into synchronisation problems. I finally wrote my own sheduler and had to give up the simultaneous threads option.
Exerpt taken from : http://www.dcs.ed.ac.uk/~fwh/emin/docs/websim/index.html
There are now several other Java based discrete simulation environments available. Like simjava they are mostly Java versions of existing simulation languages, so all implement different flavours of simulation. The advice is to use one which you're used to and which fits the application. In particular most of the below provide better direct support for queue modelling and statistics classes than Simjava. Page (1997) maintains a well structured survey of web simulation environments.
Simkit (Buss and Stork 1996,1997) provides a set of MODSIM II influenced Java classes, with no model animation but a better set of statistics classes than Simjava provides. JavaSim (Little 1997) is a text only Java version of C++SIM, itself based on SIMULA. JSIM (Nair 1997) supports a good graphical environment for displaying queues, and uses a Java database for storing results. DEVS-Java (Zeigler 1997) is a Java port of the DEVS-C++ modelling environment.
Goad (1997) describes an interesting approach to the problems of building simulations out of components, using an ML like language which unifies the behaviour and connections between components. An good example of a Java model built without using a simulation package is a model of a virtual memory system (Chakraborty and Bhowmik 1997). Digital Workshop (Lynch and Fishwick 1996) has graphical model construction for simulating simple digital circuits, as well as design animation.