Research and Modeling of Cognitive Rule of Name Novelty Based on Machine Learning and Random Forest Method

Different names make people experience differences in novelty. Names make people feel novel, while others make people feel ordinary. There is a certain pattern in people’s perception of the novelty of names. We conducted a survey questionnaire on the novelty of names based on 100 young parents and explored the novelty of 200 real names among young parents as the dependent variable for the study. Then, based on the relevant theories of name novelty research, feature extraction was performed on these 200 names as independent variables. Based on the decision tree method and the random forest method, the relationship between name novelty and name features was studied. The experimental results showed that the F1-score of the random forest model reached 85.4%, which better fitted the cognitive patterns of young parents towards name novelty but the interpretability of the random forest model is poor. The F1-score of the decision tree model reached 81.6%, which also was a high accuracy, and the interpretability of the decision tree model was strong. The decision tree model showed the “frequency of use in Chinese characters”, “frequency of use in names”, “length of names”, “number of strokes”, and “number of results on the search engine” as key variables that affect the novelty of names.