Research on Wordle Difficulty Classification Based on Multiple Model Construction

To research the classification of the difficulty for words in Wordle, this paper launched a systematic data analysis. First, considering that the "difficulty" is a qualitative indicator, this paper established a TOPSIS optimization model based on the entropy weight method to conduct a quantitative analysis of the difficulty level and obtain the score of each word. Next, by using the K-means dynamic clustering method, this paper clustered the quantitative score of the word difficulty degree, and successfully divided the words into three levels: easy, moderate and difficulty. Then, through the analysis of variance, 3 valuable attributes of the word and their relationship between the difficulty could be found. To be specific, these attributes were: whether there were two or more repeated letters in a word, whether the first letter of the word was a vowel, whether there were two consecutive identical letters in the word. Finally, the paper used the neural network prediction model to optimize the above mentioned scoring model and obtained the score of the word EERII was 0.016, meaning that the difficulty of EERII was medium. The classification model established worked well, and by means of testing, it could conform to the correlation between word difficulty grading and attributes. The difficulty classification method obtained in this paper can help Wordle game developers to assess word difficulty in time, which helps develop new game modes, adjust word difficulty to retain players, and even provide theoretical support for the sustainable development of Wordle.