| Nowadays “Micro Age” fully enter our daily live, the problem informationexplosion is the biggest problem plagued uses. Microblogging is a newly networkcommunication platform, users can freely express emotions, news commentary, etc.,but the sharp rise in the number of users bring large amounts of new informationevery day, making the user overwhelmed again. Researches on microblogginginformation aggregation is endless, but most of the service information aggregation isconcerned on the polymerization of the event itself or is associated with the events inthe same field of information, how to access useful and relevant information faster inthe chaotic and complex microblogging become an important aspect to improve userexperience.Users on the network are not interested in network resources inherent in the sizeof the amount of data, but those who can to a certain extent to meet the personalizedinformation they need, users will want to master the distribute of these information asmuch as possible in the shortest possible time. The thesis start from the user browsingbehavior based on mcroblogging behaviors characteristic analysis, according to themcroblogging information provide by user behavior data, web information extractionalgorithm and the segmentation algorithm are used. Web extraction algorithm usingthe incremental microblogging text search method, obtain information by searchingand analyzing HTML node according to web page structural features. As "Micro" newentries appear on the network frequently, in order to make the data more humane, thethesis improve the segmentation algorithm, preprocessing information based on term frequency statistics algorithm, screening out the words of high frequency for userdictionary, thus improve the correct rate of the segmentation, and to make entriesmore valuable, automatically collect and analyze user data to build the library ofmicroblogging entries relationship by incrementally searching and analyzingmicroblogging information. The thesis build the visual model to show interactive datawhen a popular topic is under discussion in the visualization part, the force-DirectedAlgorithm is used to present the straight line model, the algorithm is improved byadding the behavioral response of user actions, making the algorithm moreoperational at the same time maintaining the advantages of the layout, thus improveuser experience. A hierarchical edge bundling thinking is also used to build the circlevisualization model, the model can clearly show the relationship between the data byconstructing a quadratic curve to form a harness and reduce the burden of user visualat the same time. Users can access to information they interested faster by aggregatingmicroblogging around.In the thesis, a visualization system based on user behavior is built, start for Sinapopular microblogging topics, access the popular user-generated data in a certain timethrough Sina API, and build clouds of hot topics, automatic collect data generated byusers when some relative topics are under discussion in background operation, thenextraction and analysis the microblogging, dig to the relationship network of theentries, iterative computation and analysis of the entries, to make the data more in linewith the real. At last two visualization model are presented to the user through thevisualization techniques to help users faster access to information on specific entry,and able to predict user behavior and the rules of development of network informationbetter, also to master network information trends and monitor the networkenvironment.The visualization system is reflected in the topic under discussion by the largerange of online user behavior data, aggregate microblogging by analyzing andextracting information can help users acquiring information they needed in shortesttime, thus to improve user experience.Finally, the thesis show visual information interface based on collecting and analyzing the entry of the popular topic generated by the network, and raised theimproving points. |
PDF Full Text Request