The dichotomy between vessel-oriented research and content-oriented research suggests us a good orientaion towards goals of ITS research, where we find a lot of fruitfull results as "application of AI technology". Unfortunately, many of them are not presented in well-formulated theories. Needless to say, to elicit good knowledge from human, we had better know how the knowldege is stored or used in advance. This is the reason why we adoted the vessel-oriented approach at the first stage in ITS research projects.
In FITS project, we developed a domain-independent framework to examine what functions can be realized in domain-independent manner among those which are needed in ITS. We aimed at contributing to the advancement of AI technology applicable to education. In the student modeling module, for example, there are many tough but valuable problems to be tackled from both computational and theoretical points of view. However, these problems have not been addressed sufficiently on a well-defined computational/theoretical basis. The development of HSMIS contributes to building a computational basis for the student model module, that is, to formulate the diagnosis process and question generation process, to cope with the nonmonotonicity of a student's learning process, and so on.
Based on the "vessel-oriented research" above, we currently initiate "content-oriented" research project, say FITS/CSCL, in the area of CSCL(Computer Supported Collaborative Learning). CSCL has recently been attracting much attention of many ITS researchers. From the computer science point of view, we investigate new roles and functions of computers in CSCL aiming at promotion of computer techniques suitable for CSCL system design. We discuss ontological issues in CSCL paradigm to identify a system of basic vocabulary in terms of which we can describe activities of and communication among agents or participants in CSCL. On the basis of the ontology, we can design various computer-implemented support functions for facilitating the learning activities in CSCL. we currently concentrate on how to dynamically form a group of agents to facilitate group learning and conceptualize the communication among agents on the three issues such as (1) objectives of communication and its decision model, (2) modes of communication and its decision model, and (3) roles of learner models in CSCL.