I am currently working on visualizations of collaborative actions in Catchbob (the pervasive game platform I use to study the impacts mutual location-awareness interfaces). What I am interested in is to depict a chronological account of collaborative processes drawing on system logfiles (and the researcher's analysis in the form of messages categorizations, this packed in a XML file). Processes such as:- division of labor: the way a group of team-mates divide the work among themselves, in my case, it's mostly spatial (because the task is about exploring space) but there might be some simple roles. - the exchange of "coordination keys" among players: mutually recognized information that would enables the teammates to choose the right actions to perform so that the common goal might be reached.

It's not clear whether I would come up with one or more visualizations but the chronological view seems to be very pertinent (to see the evolution of coordination keys exchange). I might add a more spatial representation and potential statistical graph (like the evolution of players' dispersion in space, the evolution of the number of coordination devices exchanged).

As for the method, I now have a XML Schema that propose a grammar for the system logfiles and the annotations of the researcher. This allows me to transform my raw data into basic XML files that I could use to create visualizations. The point of this is to have a depiction of the activity, it might be useful for different persons: researchers (like me, interested in the analysis of collaboration), environment designers (concerned by evaluation), community managers or trainers (in case of sport analysis) or even users. However, I am less interested by the "user" category since it might need a different interface taking into account the fact that this awareness tools are used while performing the task, which is not what I am studying here.

Looking back at the CSCW/CSCL literature made me realize that most of the studies about activity visualization are concerned by social networks/sociograms. See for instance what is done at the Sociable Media Group like Visual Who (on the left here: Visual Who is a tool for visualizing the complex relationships among a large group of item: it depicts the Media Lab community, using the Lab's mailing lists ) or history flow (on the right: it provides a dynamic visualization of wiki modifications by multiple authors)

There is a tremendous mass of examples about such infoviz/informative art (a good way to find the most relevant ones is to browse infosthetics.com). What I am interested in is hence less the depictions of social networks but instead processes (like message exchange over time, role switching...). Mark Ackermann (in Ackerman M.S. and B. Starr (1995) "Social Activity Indicators for Groupware,". In Proceedings of the eight ACM symposium on User interface and software technology. pp. 159-168. Pittsburgh, PA: ACM Press.) calls this "social activity indicators", he describes 3 types of them: social network diagrams (as I explained above), notification of users' actions and activity graph/depictions. I am interested by the latter.

Here some examples I found that I think would be relevant for my purposes:

	This example is taken from Actogram, a good PC-based tool to analyse qualitative data. Quite complicated here, I like the multidimension and synthetic perspective, even though the layout is not that sexy.
	Proposed by Benford et al. (2002) “Staging and Evaluating Public Performances as an Approach to CVE”: "n this case, we are looking at a GANTT chart representation of the key scenes in chapter 1 of Avatar farm. Time runs from left to right and the different colours distinguish scenes that were occurring in different virtual worlds. The tool allows the viewer to overlay the paths of different participants through the structure. We see two participants (Role 2 and Role 3) in our example. "
	Developed by Fabien and Patrick for the portailvisualizer. It shows user's activities on a portal over time. This synthetic representation is one my favorite.
	This is taken from Paul Tennent's replay tool (the picture actually depicted 5 different windows of his tool "Clockwise from top left, the figure shows the video component handling two streams, the event series charting signal strength for each user over time, a histogram and time series showing summary information on a system property and the map showing the recorded positions of users based on GPS. Data is taken from a multi-user mobile application.", described in "Coordinated Visualisation of Video and System Log Data" by A. Morrison, P. Tennent, M. Chalmers (2006).

Why do I blog this? it's a "so what" question here for my research project. Based on this examples, I am now trying to clarify what i want to visualize and how. The idea is to have a main representation (such as the portailvisualizer version, maybe more like the Gantt Chart from the Benford paper) and different small vizualisation as in Tennent's work. I am still wondering about possible spatial representations such as:

Ideas or references are welcome!