Research on Data-driven College English Teaching Model Based on Reinforcement Learning and Virtual Reality through Online Gaming

Abstract

With the rapid development of corpus linguistics, the concept of DDL came into being. DDL (Data driven learning) makes full use of the network and corpus resources, changes the traditional teaching mode of conclusion based one-way indoctrination, creates a real language environment for learners, and reduces the phenomenon of interlanguage fossilization that is easy to occur in the process teaching method. It can effectively stimulate learners' interest in learning, cultivate their ability to learn independently and solve problems themselves, and achieve the ultimate goal of assisting English teaching. In view of this, this paper, based on the concept of DDL and taking the practice of English teaching reform as an example, designs a data-driven teaching model for college English, and discusses its specific implementation plan in the teaching process. The practical application of multi-agent deep RL (Reinforcement learning) algorithm is summarized and compared with other RL algorithms. The research shows that the algorithm in this paper improves the utilization rate of learning experience by 10.55%, thus greatly improving the learning performance. This study provides a new way for English teaching reform to improve students' ability to explore independently, and enriches the research on college English teaching models.