The context in which this study is developing is the field of CSCL, where computational methodologies comes in support of human-interaction. A computer can be seen as an agent into the system where the interaction happens, the computer is the logical partner, with limited responsiveness, limited features but still part of the system.

People communicates, exchanging messages, negotiating meanings. How does this happen? Meaning to places is socially assigned into a small group. Inference and misconceptions drives the grounding process. Some theories affirm that miscomprehension generates positive effects on collaborative learning because propels negotiation.

My initial question is: hacking the inference process of this “spatial communication”, using relevant information when there is a lack of comprehension, and providing misleading information when there is a good understanding, can result in a stranger grounding between the peers?

To answer this question we need to define a model of the communication process and of the negotiation of meaning. when this happen? How this happen? By which means? We need to find good tools for analysing the interaction, distinguishing the cognitive/meta-cognitive dimension, the social/cognitive dimension, the task communicative dimension.

One of the major point is to define essentially what does it mean “to understand” something.

The framework in which this work can develop is the context of “spatialised communication”, meaning messages which makes explicit use of the context-space to deliver the meaning.

The ideal system would be a mobile system in which every user is able to attach messages to space for different reasons: (a) orientate someone; (b) meeting someone; (c) describe a place for someone; (d) spam / make him/herself visible to the community; (e) remember something for himself.

Throughout the piling messages, the aim of this work is to find good algorithms able to grasp structures referring to the content of the messages in relation of the interaction of the users. From here, the system may guide the users interaction to achieve the research question goal.