In this paper we present a logic formalization of a model which allows us to represent, create and maintain a Learner Model for CSCL environments in a clear and adequate manner, in order to use it as the set of beliefs an agent holds about its user, supporting the effective collaboration between the learners in the community. Our model includes a representation method for the knowledge domain and the agent?s beliefs about the learner, as well as an inference system. This system allows the agent to propose the learner adequate learning tasks, considering his own interest and possibilities and those of the other members of the community through the establishment of group-based Zones of Proximal Development. The model is formalized in Answer Sets Programming, which gives an appropriate framework for dealing with non-monotonicity in disjunctive logic programs with negation.