Visual Analysis Of Spatio-Temporal Evolution Of Geographical Topics In Social Media

Twitter is a popular social media platform that records users' various check-in information.Text information implies the subject of the Tweet.Users' attention to topics has a certain geographical spatial distribution.However,these contents have considerable noise information.This noise information will greatly increase the difficulty for researchers to obtain topics of interest.It is important to quickly and accurately mine structured data from these massive amounts of Internet data.The purpose of this system design is to analyze the spatial and temporal evolution of geographic topical heat based on Twitter data in social media.Through global overview maps,topical relational views,topical flow maps,and Nightingale diagrams and other visualization tools for topics of interest The spatial and temporal evolution is analyzed and displayed from various perspectives.The main work and achievements of this paper are as follows:(1)Social media data cleaning and preprocessing.The core of data visualization is the data itself.In order to show the best visualization effect,the data needs to be preprocessed firstly.For the problem that spatial and temporal data are unevenly distributed in the ground,DBSCAN clustering algorithm is used to spatially cluster,and then Statistics on the distribution of data worldwide.(2)Map preprocessing and visualization component design.For the problem of spatio-temporal evolution of the topic in the geospatial space,the system uses the Tyson polygon to divide the global geographic location based on the data clustering through the DBSCAN space,and helps to analyze the temporal and spatial evolution of the topic through various visual components..(3)Improvements and design of visual analysis algorithms.For various practical problems encountered in clustering,this paper improves the existing algorithms to make it more suitable for the data display process of the system.For the evolution of the topic displayed on the map may lead to the overlay of the graphics,an adaptive algorithm was designed to optimize the problem and improve the visualization results.(4)Visual analysis of hot topics based on system implementation.For the problem of spatio-temporal evolution of different topics,this paper demonstrates how to use tools to analyze the spatio-temporal evolution of emergencies,using two real world data to show the evolution of topics in the spatio-temporal dimension,demonstrating the effectiveness of the system.Finally,summarize the article and look forward to the future direction of work.Based on social media data,this system uses the Elastic Search database as a data storage platform.After the geographical view divided by Voronoi,the social media topics are displayed from the local to the global,and the relevance between different topics is performed.Visual analysis,based on the improvement of existing algorithms,designs an adaptive algorithm to optimize the visual presentation process.Through the design of various interactive tools,the distribution of topics and the process of geographical evolution are explored.