In this work an Intelligent Tutoring System (ITS) is presented that is able to enrich its knowledge base, merging both partially structured and unstructured information in a unique internal representation framework. The system gathers new information about the domain in response to both a curiosity mechanism and sensory stimuli coming from the interaction with the user. The architecture is an evolution of TutorJ, an actual ITS already presented by some of the authors. The cognitive framework is a very promising one to overcome the limitations of classical ITSs, and to achieve the stated goal. The use of cognitive architectures is intended for a double result: a lower effort for a knowledge-intensive activity like the construction of an ITS and better results in terms of a rich interaction with students. In this framework the system is aware of its internal status, and a curiosity mechanism has been modeled to gather new knowledge from web sources that is integrated in the original knowledge base. The whole architecture is explained in detail, and the information fusion mechanisms are presented along with a simple case study.
展开▼
