Artificial Intelligence and Cognitive Science Case Grammar on Solving Chinese Arithmetic Word Problems in Elementary Schools

Abstract

Artificial intelligence and cognitive science technologies are combined to develop a problem-solving system of Chinese arithmetic words in elementary schools. Students can input an arithmetic word problem or update the existed problem into the system. The system automatically transforms the problems with a semantic network based on the cognitive science case grammar. In this way, the student can understand the meaning of the arithmetic word problems more clearly by the semantic network. For finding the answer, the keywords of the question part spread in the semantic network based on the spreading activation theory to find the correct answers. Students can change the question part, or add, delete, update sentences of the problem during the interaction with the system. In this way, students understand the correct meaning and knowledge of the arithmetic word problems and achieve the purpose of self-learning. For obtaining the correct label used in the case grammar for understanding the role of each Chinese word, the convolutional neural network (CNN, an artificial intelligent neural network model) is used to learn more than 600 Chinese arithmetic word problems (addition and subtraction) during the construction of the system. Experiment results showed that the system had achieved a high performance (accuracy>90%). This is an innovative technology by combining artificial intelligence and cognitive science case grammar to solve Chinese arithmetic word problems. It is also attractive and practical for online learning of other courses.