The application of intelligent algorithms in word discrimination

Abstract

Words play a huge role in people's communication and transmission of information. The LSTM model is first established in this paper to analyze the changing trend of the number of people in the time series of the data set. According to the model, the linear regression model was used to process the word characteristic values and put them into the least square model for fitting through linear regression, and the MAPE value was obtained, and the comparative test effect was conducted on the value. At the same time, the F statistic was used to test the significance of the regression equation, and the Prob value was obtained. After the comparison of standard values, word attributes did not affect the percentage of the number of people registered in the difficult mode. The characteristic values of 5 words were divided by analysis and input into LR linear regression, XGB, random forest, GA-BP neural network, and Bayesian classifier models for training. It was found that the XGB determination coefficient of the simulated annealing model was 0.506. Finally, BP neural network learning based on a genetic algorithm is used to predict, and the percentage of correct answers to each word is subject to normal distribution results.