Individualized feedback is an important factor in fostering effective learning. It is, however, often not seen in schools because providing it places a significant additional workload on teachers. One way to solve this problem is to employ critiquing systems. Critiquing systems, however, require significant development effort before they can be put into use. In this paper, we describe an incremental approach that facilitates the development of educational critiquing systems by integrating manual critiquing with critique authoring. As a result of the integration, the development of critiquing systems becomes an evolutionary process. We describe a system that we built, the Java Critiquer, as an exemplar of our model. Results from a pilot test and real-life usage of the system have shown that the system successfully provides a setting for accumulating critiques and at the same time supporting teachers in critiquing student code.

Critiquing systems check for weaknesses in the user's solution and may suggest corrections. Although their usefulness was first emphasized in diagnostic domains [6], the main success of knowledge based critiquing systems have been in design applications [4, 13]. While in the latter a deep understanding is often not necessary for critiquing purposes, in complex diagnostic domains the capability to solve the problems is critical for critiquing systems. However, they need additional knowledge to adjust to the user's solution instead of merely inferring solutions by themselves. We analyze that knowledge and propose a minimal model for extending a diagnostic consultation to a critiquing shell.

Diagnostic expert systems have long been considered an area where eventually a killer application might emerge . Much time has passed since the first prototypes were demonstrated, but we have not yet seen it in the marketplace -- despite many less spectacular success stories. Is the original idea doomed or will the technology finally live up to the expectations? In this paper we survey the state of the art with an emphasis on highlighting specific values of individual methods as well as considering the context of their use. The ultimate goal is to identify conditions and matching methods that will lead to the kind of success that pragmatist customers will find convincing - and then and only then, a real market presence will result.

Expert critiquing systems in education can support teachers in providing high quality individualized feedback to students. These systems, however, require significant development effort before they can be put into use. In this paper, we describe an incremental approach that facilitates the development of educational critiquing systems by integrating manual critiquing with critique authoring. As a result of the integration, the development of critiquing systems becomes an evolutionary process. We describe a system that we built, the Java Critiquer, as an exemplar of our model. Results from real-life usage of the system suggest benefits for supporting teachers in critiquing student code.

this article, we described three types of systems: intelligent tutoring system, standalone expert critiquing system and computer supported critiquing system. They all can provide helpful feedback to student code, but each has its own pros and cons. Tutoring systems can provide step-by-step support for completing a program, but they usually need to have extensive knowledge about the domain content, student modeling and pedagogical strategy [22]. Users in the system can only work on a predefined set of exercises. Expert critiquing systems do not have this limitation. They can perform critiquing on any code, which enables them to be beneficial for both beginner and intermediate level programmers. As with intelligent tutoring systems, expert critiquing systems require significant development effort before they can be put into use. Computer supported critiquing systems avoid this difficulty by allowing incremental authoring during real use of the system. Systems can be put into use during the early development stages. They require, however, a human in the feedback loop to ensure the quality of the feedback and handle situations the system cannot. While most systems mentioned in this article have proved effective in teaching programming, they are primarily for research purposes. Few systems have been deployed and used in a large scale. Nearly all of Page 4 of 6 Intelligent Educational Systems for Teaching Programming them were started from scratch. Reusable infrastructures with authoring tools are needed to reduce the development effort for these systems. Eventually, integration into commonly used programming development environments can make the educational experience provided by these systems more available to real-life programmers