Most traditional Chinese language teaching platforms are still based on fixed curricula and standardised progress plans; thus, they fail to meet the personalised demands of lifelong learners. Therefore, there will be problems with learning content and ability mismatches, as well as overly rapid or slow progress. Therefore, in this paper, an adaptive Chinese language learning platform has been constructed by combining a deep learning-based personalised recommendation system with natural language processing technology to address the above problems. The platform collects data and observes the behavior of learners to construct a personalised learning model, changes the learning path and content dynamically, and recommends relevant learning resources. Speech recognition and synthesis technologies are employed to provide a voice-interactive mode, and at the same time, errors in pronunciation and grammar are immediately corrected. In addition, reinforcement learning algorithms are used by the platform to automatically adjust the difficulty and type of tasks based on the learner’s real-time performance to address the problem of mismatch between learning content and progress. Based on the experimental results, after three months, the test group achieved a fluency score exceeding 3.9 and reduced the number of pauses in oral expression to less than 10; thus, personalised learning was improved. The platform provides good, flexible learning support for lifelong learners; it has practical value and helps promote the development of intelligent language education.
