Knowledge Control System with the Possibility of Fee-Constructed Answers Recognition

This magister’s dissertation covers main aspects of designing and developing an automated knowledge control system with the ability of grading freely constructed answers. As parts of undertaken system analysis, task decomposition, domain knowledge analysis and modelling of the system to be developed were committed. This paper touches on the topicality of using automated computational system in education in general and using automated knowledge control systems in particular. The dissertation compares existing knowledge control systems and points out their common shortcomings, for some of which means of overcoming are suggested. Different types of questions used in modern tests were analyzed and classified. Methods of automated grading of such questions were subjected to analysis as well. For most common question types algorithms of automated grading were proposed, including methods of natural language processing, highly structured text processing and image recognition. Considering results of the research made, a prototype of knowledge control system with the ability of grading freely constructed answers was created. The system was tested and proven to meet imposed requirements. During the final step of dissertation, conclusions were made based results of the research and the following paper was written.