| With the popularization of computers and the Internet,computer input has replaced people’s habit of writing with a pen,promoted the development of text storage in electronic form,and provided conditions for the application of text mining in literature.In the study of literary style,by using the method of text mining,it not only makes up for the insufficiency of the previous introspective research,but also creates a new idea for the study of writing style.As representatives of the avant-garde,Ge Fei and Yu Hua’s novels present a style completely different from traditional literature.By studying Ge Fei and Yu Hua’s novels,the field of style research can be broadened.At present,using the method of text mining,there are few comparative studies on their novel styles,and there is a large research space.This article selects 37 novels by Ge Fei and Yu Hua.After text cleaning and word segmentation,a corpus of Ge Fei and Yu Hua’s novels is constructed.On the basis of constructing corpus and previous research,by using language feature model and bag-ofwords model to convert text data into structured data,two sets of data sets are generated.Based on the data set of language feature model,this paper selects the XGB model and random forest with good performance by using random forest,XGB model and support vector machine for prediction,and then based on these two models,five important languages are obtained.feature.On the other hand,based on the bag-of-words model data set,through the selection of model parameters,the results of the weighted LDA topic model are obtained,and then the differences in the modification styles of the two writers are analyzed.The main conclusions of this paper are as follows: First,the selected measurement features are reasonable.Ten measurement features were selected based on the language feature model,and predicted by random forest,XGB model and support vector machine,which performed well on the test set-the accuracy,recall,precision and F value were all above 0.77,especially All indicators of the XGB model are above 0.87.These indicate that the ten features selected in this paper contain information to distinguish the styles of Ge Fei and Yu Hua’s works,reflect the rationality of the selected measurement features in describing the style of works,and make up for the lack of rationality in the selection of measurement features in previous studies.of insufficiency.Second,a macro overview of the style of Ge Fei and Yu Hua’s novels.Compared with Ge Fei,Yu Hua uses colons more frequently,and his sentences are more colloquial.The above conclusions are consistent with the previous research conclusions,but the previous research focused on some novels,and this paper uses most of the author’s works for research,which broadens the scope of application of the previous conclusions to a certain extent.Different from previous studies,this paper finds that Yu Hua’s novels tend to use a lot of dialogue descriptions to portray characters.In the use of modifiers,Ge Fei mostly uses adjectives with negative emotions,and Yu Hua mostly uses adjectives with positive emotions. |
