Research on Natural Language Recognition based on Grey Correlation Degree and TF-IDF Algorithm

Natural language research topic in recent years is complying with the important topic of artificial intelligence development, when dealing with text information data, the training mechanism and combined with the feature of algorithm for natural language data analysis is an effective method to deal with complex text information. For the text classification and judgment process, in combination with the concept of deep learning algorithms have been proposed, and get the argument, combined with the feature of countable characteristic of text classification problem. In the process of research, it often need a lot of samples, the complexity of the samples affect the key content of text extraction, but the lack of samples will lead to extract text content and it is too simplified, the deviation is from the characteristics of information content, On the other hand, for the huge amounts of text sample, it will lead to more redundancy in the compilation process of the algorithm, which has certain requirements for the configuration of the hardware system, and it loses timeliness in the operation time. In this paper, the accuracy of feature extraction of complex text is studied by using an innovative idea of algorithm iteration, and the mathematical grey correlation degree is used to deduce and verify the accuracy.