Development Trend of Code Defect Detection Technology Based on Natural Language Processing

Traditional defect detection research is based on program analysis, and when dealing with large-scale programs, the detection speed is usually slow, resulting in limited detection performance. Moreover, because traditional defect detection can only extract basic semantic information, the detection accuracy is not high, which makes it difficult to play a role in practical applications. In recent years, with the progress of artificial intelligence technology, natural language processing technology has also been rapidly developed. Through natural language processing technology, more complex, highdimensional and multi-level semantic features can be extracted from the code, so that the code defect detection has a higher accuracy, and the detection speed has been significantly improved. In this paper, the author mainly collects the relevant literature of code defect detection based on CNN, RNN, LSTM and Transformer, which belongs to natural language processing technology, discusses the difficulties and challenges still existing in defect detection, and finally makes a summary and outlook for the future.