Stories are an ancient genre of art and literature,while data stories are a new technique of science and technology.As a brand-new application area,data stories have attracted much attention in industry,but the academic community is also in dire need of groundbreaking research on its key issues.The formulation of the definition of data stories unifies the differences among existing definitions of data stories,focuses on the key contradictions in data storytelling,deepens the understanding of data stories,and better supports the automatic generation of data stories.At the same time,the proposal of the two-stage theory of data storytelling not only further explores the definition of data storytelling but also decomposes data storytelling into two relatively independent activities of generation and description.Usually,there are three basic scientific questions behind data stories and data storytelling:what-if questions,why-not questions,and how-to questions,which focus on exploratory analysis,explanatory analysis,and instructive analysis,respectively.Data stories have four main features,namely data,story,business and science.These four features of data stories are proposed to better describe the differences between data stories and literary stories and data visualization works.A data story usually consists of five elements,namely the character,the event,the plot,the data insight and the business purpose.The proposal of the above five elements of data stories has eliminated the confusion of sources and components in previous research,corrected the misunderstanding of the one-sided emphasis on the status of data visualization in data storytelling,and realized the creation and narration of data stories.The DAIS model for data story generation clarifies not only the four key stages of the data story generation process—data,analysis,insights and story—but also the work content and procedures in each stage.From a methodological perspective,data story generation methods can be divided into four types:model-agnostic global storytelling,model-specific local storytelling,model-agnostic global storytelling,and model-specific local storytelling.Data storytelling has three main functions:to experience,to explain,and to enlighten.Data storytelling is currently used primarily in data analysis and model interpretation,metaverse application and linking virtual and real worlds,teaching and training,brand advertising and digital marketing,data-driven management and decision making,and content creation and product design.Data stories bridge the real and virtual worlds,and the exploration of virtual-real synthetic data stories is becoming an important topic in the study of the metaverse.8 figs.3 tabs.58 refs.
Data storyData scienceData insightData analysisNarrative
NETL
NSTL
万方数据
