Design And Implementation Of Sentiment Analysisi System For Microblog Short Text

The advent of 5G accelerate the speed of information transmission.With the era of big data age is coming now,various software has influx into people's lives.Software users have changed from passively receiving and downloading information to now actively creating and uploading information.Weibo has a major position in information circulation and has become more and more popular,because it not only able to acquire knowledge quickly,but also express own opinions for freedom.It's meaningful for extracting people's attitudes from massive information,understanding user values and emotional positions,which help the government to grasp the public's tendency,monitor public opinion,formulate relevant policies,improve user's satisfaction and commercial profits at the same time.The thesis narrates from the following aspects.First,investigates and analyzes the advantages and disadvantages of commonly used text sentiment analysis algorithms.Second,analyzes the sentiment analysis tasks at the sentence level for the short text characteristics of Weibo microblogs.Third,adds attention mechanism on the Long Short-term Memory Network.After pre-training with Word2vec,the word vector whose data dimension is reduced is input into the long and short-term memory network based on the attention machine.The attention mechanism can directly learn the emotional information of each word in the word vector,and generate corresponding word vector weights.Through sum the weighted words,we can get the basic sentiment polarity of the sentence.The paper constructs a corpus using a 5 million microblog corpus provided by the Beijing Institute of Technology's search and mining laboratory,and compares it with different models of SVM,LDA,LSTM,ATT+LSTM.The long-term and short-term memory network based on the attention mechanism has achieved good results,which confirming the applicability of the algorithm proposed in this article in the field of short text on Weibo.On this basis of above,this paper designs and develops a sentiment analysis web application system for micro-blog short texts in strict accordance with the software engineering development process.The front end uses the Vue.js framework to design a user-friendly interface,and the back end uses the Django framework to implement interactive logic.The system collects Sian Weibo data and trains with the algorithm model proposed in the article,and realizes the functions of personal emotion classification and event public opinion prediction,which not only guarantees the reliability of the system's emotional analysis,but also enables the algorithm to successfully land.Finally,perform functional and non-functional tests on the system.