Research On Visualization Method And System Realization Of Multi-dimensional Twitter Information

With the rapid development of Internet technology and the unprecedented prosperity of mobile communication technology applications,social media has become an important tool for people to acquire and share information in social life.Users are more and more interested in making speeches and opinions on social networks.Social media also benefited from the growth of users,and thus played an extremely important role in many important traditional areas.The important role of Twitter in the event of political events,natural disasters and other events has made great attention to event detection and event visualization of social networks.Twitter event visualization refers to the process of displaying event in the graphical interface and providing the appropriate event information process.In the visualization of events,the location plays an important role in assessing the impact of events and the development of perceived events.Therefore,identifying the geographical location of the events is a prerequisite for visualizing the events.Multiple grams refer to noun phrases that represent complete names,place names,and so on.Statistics found that more than half of the names are multiple grams,in order to accurately locate the event,there is need for multiple grams identification.This thesis focus on the social media short text of the multi grams recognition technology,event positioning technology and event information visualization method.This thesis mainly in the following aspects of the innovative design and implementation work:(1)A multi grams recognition method based on unsupervised learning is proposed.In order to realize the multi grams recognition,a preprocessing method based on pattern matching is proposed,and the tweets are normalized and segmented by matching features.Based on preprocessing,an unsupervised multi grams recognition method is used to extract multi grams information in the non-annotated set of tweets.In the actual tweaking training and testing,compared with the traditional multi grams recognition method,this method can significantly improve the accuracy of multi grams recognition with the increase of the number of processing tweets.(2)The method of positioning the Twitter event based on the maximum entropy model is proposed.This method improves the feature template of the maximum entropy model,constructs the feature with multiple semantic information,and classifies the maximum entropy model using the lexical annotation and the training set marked by the entity.A weighted location algorithm based on location information is proposed to locate the Twitter events according to the geographical location of the text,the user position and the push position.This method improves the accuracy and recall rate of calligraphy recognition.At the same time,the event location algorithm can locate the event efficiently and accurately.(3)Design and implemente a set of geographic information fusion event visualization system.The system is based on the location of the Twitter event information visualization system to achieve the realization of the event pre-processing,positioning and real-time visualization function.The system is based on geographic location information,and presents the event information with the electronic map.In the actual test,the system can be stable to complete the visual work.