Studying misconceptions, researchers often face a choice between getting insight into students' thinking (using free text interviews) and the number of processed answers (which affects the study's validity). In this poster, we propose a technique of analysis students' answers at free-text questions requiring interpreting program trace lines to study misconceptions of control flow statements. It allows labeling free-text answers for quantitative study and processing them more efficiently than during unstructured interviews while still getting free expressions of students' thoughts.

The volume of information that can be used in the development of knowledge bases that can be used in education is constantly increasing. Also, this amount of data is very difficult to process and store. When designing a knowledge base to optimize the educational process, it is important to use ontologies. At the moment, the creation of an ontological knowledge model is the most promising option for storing and processing information. The article describes effective approaches for generating an ontological model using machine learning models based on the Transformer model.

When developing automatic quizzing systems and intelligent tutoring systems, significant effort has to be spent on developing the problem bank. Question and problem generation is a field of study concerning automating this routine work. The two most common methods of problem generation are constrained random generation and template-based generation, but each of them has disadvantages. In this work, we study a possibility of generating code-tracing learning problems from existing open-source code. We generated a set of learning problems and evaluated their distinctness from human-authored problems and readiness for usage in the learning process. Both teachers and students showed the rate of determining machine-generated problems only slightly above random guessing. Teachers strongly agreed with the problems' relevance and agreed with their suitability for the learning process. Automatic labeling to filter desired problems for the assignment includes used concepts, possible errors during solving, and difficulty estimates. The studied type of learning problem required little additional data to add to the code; our further work will concern problem types with more dynamic data to overcome this limitation.

One of the typical problems in question generation is the low cognitive level of the generated questions. The problem generation field seeks to generate higher-level learning problems using either constraint-based random generation or template-based generation. Both strategies have drawbacks and often require human participation. However, learning problems on high cognitive levels often have real-world equivalents that makes problem mining a viable approach in various fields of learning. The main challenge to overcome in problem mining is problem classification: automatic evaluation of the parameters of the problem’s pedagogical usage. Automatically generated and classified problems can be used in adaptive tutoring systems without further work or with minimal human supervision. The paper discusses examples of possible fields for problem mining, the advantages and challenges of this approach.

When developing automated assessments and intelligent tutoring systems, a lot of routine effort goes into developing the bank of learning problems. Problem generation is the way to automate this process. In this paper, we present a method of generating expression-related problems for teaching introductory programming courses. The problems are generated from open-source software code which allows keeping learning problems similar to the production code the students should learn to analyze and write. Generated problems are automatically classified by their difficulties and the knowledge they need to solve, represented as sets of possible errors. This allows seamless integration with adaptive learning algorithms. The evaluation showed that the generated problems are indistinguishable from human-authored problems and suitable for use in the educational process.

Adaptive learning exercises intended for mastery learning are performed until the learner reaches the intended mastery level. But how to grade them if all passing students reach the same mastery at the end? Our goal was to calculate grades from the number of wrong steps and correct steps while solving the last questions in the exercise. This approach is implemented in an intelligent tutoring system and its behavior was observed during the system's evaluation. The faster the students learned, the higher were their grades. This technique can be used to grade students directly using the results of solving problems in the training system without separate summative tests. This lets students receive feedback on how well they are learning without performing more exercises.

We developed a web-based tool for learning the order of evaluating expressions in C++ and Python languages. The variety of operator precedence and associativity among programming languages and the lack of direct visualization make understanding expression evaluation difficult for some students. The key feature of the new system is a detailed explanation of errors, containing fault reasons---the subject domain laws that the student violated. We evaluated the tool with 14 first-year Computer Science students and received positive feedback. This tool can be used for learning new concepts during homework without requiring more class time because it provides enough feedback for students to learn on their own.

Teaching students to construct correct statements in programming languages is an important part of introductory programming courses. While modern development environments highlight syntax errors, they do not stimulate thinking in grammatical terms to help the student understand and memorize syntax rules. To facilitate learning syntax, we developed CorrectWriting, a question-type plug-in for the popular LMS Moodle. It finds mistakes in token order and composition and detects typos, including missing and extraneous separators. Mistake messages use teacher-supplied token descriptions to show the grammatical role of each wrong token. Hints are provided about students' mistakes. CorrectWriting questions are actively used by the students of Volgograd State Technical University to prepare for classwork.

Developing an understanding of control structures is one of the important tasks in introductory programming courses. To facilitate active learning with immediate feedback, we developed a constraint-based tutor, "How it Works: Algorithms," that asks students to build an execution trace of the given algorithm and provides explanatory feedback about the mistakes the student made. The reasons for the student's faults are determined by an inference engine, using a set of rules describing the subject domain. Teachers can create exercises using a simple visual block-based interface and sending them to students as permanent links.

One of the ways to demonstrate subject-domain concepts is to let the learner play in the domain-related sandbox, while receiving the explanations of the domain laws that were broken when the user makes a wrong move. This can help the learners who understand definitions of the concepts poorly. We present two visual simulators for demonstrating concept properties and domain laws, implemented as web applications. They can be used to study programming, enabling trial-and-error learning, supported by the error messages, explaining why the user's action was wrong.