QUESTION GENERATION FROM KNOWLEDGE GRAPH WITH PRESCRIBED DIFFICULTY LEVEL

Abstract

Automatic question generation in Knowledge Graphs (KG) is a novel idea. KG is the graphical representation of knowledge bases where knowledge base is world knowledge. Intelligent Tutoring System (ITS) is a computer program that uses artificial intelligence techniques to improve learning process in the knowledge domain. An ITS provides adaptive learning instruction and feedback to students in problem solving process. KG can be used in Intelligent Tutoring Systems for education purposes. In this research we address the problem of automatically generating knowledge questions from a KG with assigned difficulty level (easy and hard). Questions of this kind have ample applications, for instance, in Intelligent Tutoring Systems, to evaluate the knowledge of students in a specific domain. Automatic question generation can reduce human effort as making questions manually is resource and time consuming. To solve the problem, we propose a novel approach of generating questions automatically. To generate questions, we first select a domain for the knowledge graph; then process the KG to generate structured triple-pattern triples, which are then used to generate questions. A key challenge is estimating how to generate easy and difficult questions. To do this, we used single triple for easy questions and more than one triple for difficult questions which means that using only one triple involves two nodes and more than one triple contains more than two nodes, so when the graph is traversed for easy questions only two nodes are visited and for difficult questions more than two nodes are visited which increases the complexity.