Ph.D august meeting: model, mutual model and schedule for 2005-2006
This morning, I had a meeting with my Ph.D advisor to set the boundaries of my research project. Apart from the experimental aspects of the project I already done (the CatchBob! project), Pierre wants me to describe a formal model (the thesis would be defended in computer science, that's why) of the phenomenon we study, namely how Mutual Modeling is deployed to support coordination in spatial task. The use of such a model is to help designers to build better interfaces/analyse data or logfiles and so forth for applications devoted to mobile collaboration (games, firefighters thing...). As for the form of the model, the use of a formal language/data structure (UML, XML, something-ML...) or a graph/table is required. The point is to build this model based on the results of my first (and ongoing) experiment. The second experiment would be designed to illustrate/validate/deploy this model (or to change it).
How to move forward:
- from the first experiment, list all the 'coordination devices/keys' (Clark's theory of coordination) and link them to performance/mutual modeling
- from the interviews/confrontation to the replay tool, distinguish which information are quoted by the users, in terms of: Group Model, Self Model, Mutual Model_A(B), Mutual Model_A(C), the representation of the environment. The idea is to describe the important information used by agents to coordinate.
- this should be based also on Clark's theory of coordination, as well as Sperber and Wilson theory of mutual cognitive environment.
- september/october: model description/formalization
- THEN: pick up the variables we want to test in the second experiment
- Second experiment: january-march 2006 BUT it's going to depend upon whether we keep the same environment (and we add some more complex features) or if we need to have a new one.
How to more forward in the 'mutual modeling' project:
- does the teams with a good MM give more information in terms of the mutual modeling information they used?
- check the interaction asymetry and the quality of mutual modeling (correlation + intraclass correlation)
- in catchbob: list all my variables: useful for corelation/used as coavriables/regression analysis!