A Study On Personalized Teaching Assistant System Based On Knowledge Points And Relationships

At present, establishing domain knowledge in computer assisted instruction system becomes rather difficult, which both need heavy workloads and collaboration of experts in all aspects. Furthermore, the individuating demands of students always can be satisfied. In the process of collecting students'information, it can not take students'daily performances into account systematically and comprehensively.In this thesis, we propose a knowledge-centered computer-aided instruction system which for personalized programming teaching. The main contents of this thesis include:(1) Natural Language Processing techniques are employed for the knowledge points extracting. This paper uses a unitary statistical model to finish the sub-word process and obtain the knowledge point of the candidate subset based upon the segmentation unit maximum matching algorithm. (2) Three knowledge-based database models are designed which closely related to the teaching system.(3) Then, personalized learning contents are generated in the actual teaching process by implementing individual student information database in accordance with test database annihilator map.(4) The cognitive capability level of students are classified by fuzzy set theory; According to Kolb's leaning theory of classification, the study styles of students are extracted; The examination questions are mapped to knowledge points by the way of reverse maximum matching.(5) An Knowledge-Points based Instruction assistant system are implemented based on Browser/Server structure which can be accessed from web.The characteristics and innovations of this paper are(1) The relationships of knowledge points are extracted by a novel way which contains three parts that are a improved hybrid text classification algorithm and derivation rules in artificial intelligence and association rules in data mining.(2) The individuation sub-graphs of students'information on mastered knowledge are extracted by the ideas of directed graph.(3) Both cognitive capability level and study styles of students are incorporated into teaching systems, and the individual learning contents are extracted by a dynamic generation system, then, more comprehensive individual students' information is manifested into the students information database.Experiments show that the accuracy and recall rates of extracting relationship between knowledge points have been improved during the teaching process. Therefore, KrePerS system can play a good supporting role in the learning process of students.