Conversational intelligent tutoring systems (CITS) use natural language dialogue between the learner and the system as the medium of instruction. In the context of a framework for CITS called Oscar, this paper discusses personalization of instruction based on learning style and other parameters. A tutoring system for structured query language (SQL) is discussed as a case study for the framework. The learning style of a learner is determined by observing his interaction with the system and his behavior. The focus is on an adaptation algorithm that attempts a best match between available instructional resources and resources that suit the learner. Empirical studies using the SQL tutor with 72 students are reported showing that the SQL tutor improves learner performance. However, in the absence of a direct comparison to any other adaptation algorithm, it is hard to accept claims on the impact of the proposed adaptation algorithm.

The paper starts by providing an overview of the field and discussing the context of the paper. Section 2 provides a brief review of ITS, conversational agents, and learning styles. This is followed by a discussion of Oscar and the adaptation algorithm. Section 4 discusses Oscar for SQL and the proposed adaptation algorithm; an appendix at the end gives more details on the algorithm. The empirical evaluation of the claims is discussed in section 5. Some thoughts on generalizing Oscar are presented in section 6, with section 7 concluding the paper.

Overall, it is a good read and contains a fair amount of useful information. It will be useful for people with an interest in CITS.