Research On News Classification And Recommendation Method Of Taiyuan Education Bureau Government Affairs Big Data Platform

With the rapid development of big data,cloud computing and artificial intelligence,the traditional E-government has been difficult to adapt to the new situation.Building a government platform integrating big data,digitalization and intelligence has become a new way for the government to seek development.Since 2017,the construction of big data platform for government affairs of Taiyuan Education Bureau has made continuous achievements,but there is still a large gap compared with the government affairs platforms of developed countries and advanced provinces and cities in China.At present,the news classification method adopted by the platform has coarse classification granularity and low classification accuracy and efficiency.At present,the recommendation methods used in the platform can not reflect the changes of users' interests in time,the accuracy of recommendation results is low and the degree of recommendation personalization is low.This paper studies these problems in the government big data platform.In this paper,the traditional TF-IDF feature extraction method is improved,and a new feature extraction method,ETF-IDF-L,is proposed,which combines the education government vocabulary and the location information of feature words The feature extraction method is used to classify the educational government news by naive Bayesian network and convolution neural network respectively.The experiment shows that for some categories with less data,it is based on ETF-IDF-L The result of feature extraction and vector space model text representation combined with polynomial naive Bayesian classification model is better.For the category with more data,the result of convolutional neural network classification model based on Word2 vec is better.The accuracy and precision of these two classification models are improved compared with the existing news classification methods of the platform.This paper solves the problems of coarse granularity and low accuracy of the existing news classification methods on the government big data platform of Taiyuan Education Bureau.In this paper,the collaborative filtering algorithm is improved,and a filling method that combines the average score of users with the current news popularity is proposed,which solves the data sparsity problem of the traditional collaborative filtering algorithm.Then,the improved collaborative filtering algorithm and the content-based and user attribute based recommendation algorithm are combined to build a hybrid strategy based education government news recommender Method: compared with the single recommendation algorithm,the hybrid recommendation method has different degrees of improvement in accuracy,recall rate,value and so on,which meets the user's interest change and personalized needs,and solves the user's personalized and low recommendation accuracy problems of the existing news recommendation methods on the government big data platform of Taiyuan Education Bureau.