Seq2Seq Text Recognition Method for Large-Scale Corpus Linguistics Knowledge Based on Transformer

To improve the accuracy of large-scale text recognition and reduce recognition time, this paper proposes a Transformer-based large-scale corpus linguistic knowledge Seq2Seq text recognition method. This method first extracts linguistic knowledge Seq2Seq text features from a corpus based on frequency cross-entropy, then represents word vectors based on a fuzzy word model, and finally introduces the Transformer decoding concept to construct a large-scale corpus linguistic knowledge Seq2Seq text recognition model, achieving linguistic knowledge Seq2Seq text recognition function. The experimental results show that the proposed method has a recognition accuracy higher than 93.5%, and the recognition time for thousands of pending data is less than 3.2[Formula: see text]ms, which is better than the comparative method. The application effect is good.