| With the rapid development of social networks,social platforms have become an important place for the occurrence,development and evolution of public sentiments.Therefore,analyzing the trend of online public opinion for social platforms and mining the public opinion deep behind can help managers analyze users' preferences,grasp user needs and enhance the user experience of personalization services.On the other side,it can objectively reflect the truth of the society that guided by public opinion.Our system fully reflects the social values of netizens and helps decision-makers improve their ability to deal with emergencies.Aiming at the existing online public opinion analysis system,there are problems such as hard in gathering public opinion data and analysis method too single.This article redesigned an efficient data crawler,combined with a variety of text-based sentiment analysis methods,as well as the online public opinion analysis method in communication.Then we developed an online public opinion analysis system for the Weibo platform.By constructing the system,we can get a direct sense of the dynamic and rapid show of the sentiment situation of Weibo platform,which can effectively improve the timeliness and convenience of monitoring and mastering the trend of the target public opinion development.The specific work mainly includes the following aspects:1.At system implementation level,a network public opinion analysis system based on the Weibo platform was designed and developed.The system mainly includes a five-layer structure: data collection layer,data storage layer,business logic layer,service layer,and presentation layer.It in charge of the collection and storage of public opinion information,a complete process of designing and implementing functions.Compared with the traditional public opinion analysis system,our work has a higher degree of structuring and a more detailed and clear division of functions;2.In the system data collection layer,a new type of distributed web crawler was designed and developed to collect public opinion data generated by users on Sina Weibo.By improving the traditional crawler from distributed design,downloader and parser interaction strategy,web parsing mode,etc.,it has improved the existing crawler data crawl information incomplete,unclear structure and poor timeliness and other issues;3.At the logic level of the system,we have research and use TF-IDF algorithm,sentiment polarity judgment method based on emotion dictionary,naive Bayes-based sentiment analysis method,etc.Through a keyword extraction procedure for the posts published by users,emotional polarity analysis,and text clustering,we can achieve functions like mining users' point of view,discover the public opinion and warning target timely and other functions;4.At the system protection level,system function implementation is tested from three levels: unit testing,integration testing,and system testing.Performance is evaluated using six indicators,including response events,resource usage,and the number of hits.Integrity and system effectiveness ensure the high efficiency and stability of the system. |
