Visual Analysis Of Scientific Publications Based On Image Collections

Scientific publications as the carrier for recording technological knowledge and authors' research experience,has great value to transfer the subject knowledge.With the rapid increase of scientific publication,analyzing and mining information has become a huge challenge.The visual analysis provides more and more visual analysis publication research to provide us unprecedented insights,such as literature review and collaborator network analysis.However,visual information that is widely used for storytelling and method descriptions is ignored.Therefore,this article develops a data visualization analysis system VIStory.By encoding the picture information in scientific publication,a new document visualization interactive storyboard is created to help people quickly view the visual information in scientific publications.It supports the exploration and discovery of the research content in scientific publication,the author's research area,the number of cited references,visual display of the experimental results of the paper,and efforts to promote the dissemination of data visualization research.As a tentative exploration of the research on the visualization of scientific publication sets,this paper selects two data sets.One is the paper data of IEEE VIS,and the other is the multimedia integration technology research of Shenzhen Institute of Advanced Technology,Chinese Academy of Sciences Proceedings of the Center Group.The research mainly involves the following points: using a web crawler tool to complete data collection;automated PDF extraction tools to complete extracting key information from the original paper data,mainly including the paper images,various attributes of the picture such as aspect ratio,domain color,etc.The meta attributes of the paper include the author of the paper,the conference category,the keywords of the paper,the number of cited papers,etc.;platform data modeling and database construction;platform visual interface design and interactive operation design;front-end visual display and layout algorithm implementation;And several case studies such as the visualization analysis of IEEE VIS mainland scholars based on this visualization system.VIStory visual analysis system mainly includes the following components:(1)Faceted View + Endgame View,as an intuitive data selection and detailed exploration panel.(2)Storyboard View,the layout of the ThemeRiver to metaphorize the evolution of data in a paper over time,while adding a well-designed glyph Paper Ring,to metaphor a paper Multiple data information is encoded in the paper circle,such as the number of pictures in the paper,the main color information of the picture,the picture aspect ratio,and the order of the pictures in the paper.(3)Interactive design of the platform.The VIStory visual analysis platform uses three types of interaction methods: mouse hover,click,and drag.First,the mouse hover operation is used to display the detailed panel view.Secondly,the click operation is used to select a data set and select a grouping method for displaying data.The drag operation is used to fix the current detailed panel so that users can compare and analyze different pictures.This article discusses the future development of visual information visualization in publications,put forward the important value of images in the scientific publication.In the context of the current explosive growth of multimedia data and visual information data,we proposed more attempts and explorations on visual information visualization.