Classification of Messages in ＂Smart Government Affairs＂

Zuyu Hou

《Frontiers in Economics and Management》 3卷1期 (2022/01) Pp. 429-444

https://doi.org/10.6981/FEM.202201_3(1).0051

摘要

In recent years, the online political platform has gradually become an important channel for the government to listen to the voices of the people, solve their worries, gather their wisdom, and warm the people's hearts. The amount of text data on various social conditions, public opinions, policies and people's livelihood has continued to rise. The work of related departments and related hot spots has brought huge challenges. Based on natural language processing technology and text mining methods, this paper constructs a targeted ＂smart government system＂ for the classification of public messages in the context of this political trend, and becomes a bridge connecting the government and the people. For the mass message information data set given in a certain format, one-hot encoding is performed on the training set and the verification set under the 5-fold hierarchical cross-validation division, and the machine learning method based on the TF*IDF bag-of-words vector feature engineering and the Convolutional neural network CNN, recurrent neural network RNN and deep learning methods under long and short memory network LSTM are used for message classification. After selecting the best comparison, this paper finally uses the support vector machine classifier based on the part of speech level TF*IDF model under machine learning to classify. After this step, the average value of F1-Score under 5-fold hierarchical cross-validation reaches 0.9013, and the output The corresponding predicted label data set. In the end, the automatic division of the procedures for the masses' messages was realized, which greatly improved the efficiency of dividing the masses' messages in government work.

關鍵字

TF*IDF ； Machine Learning ； Convolutional Neural Network ； Recurrent Neural Network ； Long Short-Term Memory

參考文獻

Yoon Kim.Convolutional Neural Networks for Sentence Classification[J].2014.

Nal Kalchbrenner,Edward Grefenstette,Phil Blunsom.A Convolutional Neural Network for Modelling Sentences.[J].2014.

Ye Zhang,Byron Wallace.A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification.[J].2015.

Chunting Zhou,Chonglin Sun,Zhiyuan Liu,Francis C.M.Lau.A C-LSTM Neural Network for Text Classification.[J].2015.