AI-based Learning Aids for Solving Kinematics Word Problems

E-learning is the idea of using electronic technologies for learning. E-learning has gained prominence among students all over the world over the past few years. This has triggered widespread research in the development of techniques that can aid the different stakeholders - students, instructors, of an e-learning system. The aim is to eventually have a system that all the stakeholders find convenient to interact with. In this thesis, our goal is to develop Artificial Intelligence (AI)-based learning aids for solving word problems from the domain of kinematics. Word problems or story problems are exercises which describe a real-life scenario in the form of text and follows up this narration with a query. Answering this query requires mapping the facts given in the form of text to a mathematical notation and then applying some mathematical or physics concepts. An e-learning system offering a course on word problem solving is more student- friendly if it has the ability to provide a step-wise solution to word problems that a student finds hard to solve. The first two goals of this thesis is towards realizing such a feature for the domain of kinematics word problem solving. Kinematics is the study of motion without considering the cause of motion. We develop a system that can automatically solve word problems from kinematics domain. Such a domain specific word problem solver has to first extract the numerical facts required to solve the problem and then reason with the domain's knowledge to infer the procedure to be applied to solve a given problem. The procedure, here, refers to the steps involved in computing the value of the unknown quantity in the problem. The student using the solver expects an explanation as to why a certain step in the solution is important.