Abstract
The work was centered on observing actual computer scientists communicating on the web via social tools (instant messaging and video conferencing) in the context of a joint project. The objective has been to understand how learning and communication mutually influence one another; allowing people to infer each other's communicative behavior, at the same time understanding how intentions arise when people are speaking and doing activities. First, actual conversations have been recorded and observed (about 50,000 words exchanges). Then those conversations have been manually translated into formalized agent messages. The analysis of those formalized messages, and the comprehension of the communicative scenarios has required the extensive application of existing theories: (i) hierarchy of learning and communication (i.e. logical theories of learning and communications) (Bateson, 1972 & 1979); (ii) situated cognition (Clancey, 1997); and (iii) activity theory (Leont'ev 1977 & 1978). However, during this analytical process, an innovative theoretical framework has emerged that has been called Activity States. It provides guidelines on how to convert the actual conversations into agent communication messages (having equivalent semantics). The Activity State framework also attempts to understand and explain how the activity of reading (as an example), and comprehending the text that one reads, is in relationship to that person's activity on the web. All of this influences how people formulate his/her intentions. Finally, the analysis of the formalized messages enabled preliminary findings: (i)people have internal rules (e.g., a combinatorial rule system); and (ii)people learn, merge and adapt communication protocols in their situated context (in some ways validating some existing theories suggesting this). As a conclusion, our Activity States framework is claimed to be a promising approach for a better understanding of human collaborative behavior at a distance, over the Web.