| Beijing Forestry University mainly relies on the Green News website to release school-related news,In order to keep up with the development of The Times and facilitate users to browse news on mobile terminals,the Publicity Department of Beijing Forestry University urgently needs to build a mobile platform.The platform can support Green News website to manage and publish school-related news on mobile terminals,and support user comments and corresponding sentiment analysis.Based on above requirements,the BJFU Green News Mobile Platform,including the mobile news management subsystem and the wechat mini program subsystem,is designed and constructed in this thesis.A multi-model fusion model is also designed and realized to achieve automatic sentiment analysis of BJFU News comments.The main work of this paper is as follows:(1)Aiming at the needs of sentiment analysis for news comments,a multi-model fusion model is designed and realized..In this model,Bert pre-training model is used to initialize the comment characterization,and multi-model with Res Net,Capsnet and Muilt-Bi LSTM is used to extract and fusion the sentiment features to get the polarity analysis of the news comments.The experiment results on the dataset of BJFU campus news comment and public Dianping dataset show that the model has good performance and can satisfy the sentiment analysis needs of Green News website.(2)According to the actual demand of BJFU Publicity Department to pick some news from Green News website and release them on mobile terminal,we design and construct a web based green news mobile management subsystem.This subsystem is constructed by the development framework‘Spring Boot+Vue’ and provides functions such as mobile news releasing/removing 、 news comment managing and comment sentiment analysis etc.(3)To satisfy needs of mobile readers of Green News website,a wechat mini-program subsystem of Green News website is designed and developed.By this wechat mini-program,users can browse different types of news,comment on the news and collect favorite news etc. |
PDF Full Text Request