Supporting Story Synthesis: Bridging the Gap between Visual Analytics and Storytelling Siming Chen, Jie Li, Gennady Andrienko, Natalia Andrienko, Yun Wang, Phong H. Nguyen, and Cagatay Turkay (2018) 1
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 2
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 3
Visual Analytics Figure 6. Visual analytics interface of social media, including (a) topic projection view, (b) topic comparison view, (c) user/ location distribution view, (d) temporal view, (e) story slice generation parameters, and (f) raw data table. 4
Visual Analytics “Fancy visuals, cool interactions, but what does this mean?” Figure 6. Visual analytics interface of social media, including (a) topic projection view, (b) topic comparison view, (c) user/ location distribution view, (d) temporal view, (e) story slice generation parameters, and (f) raw data table. 5
Storytelling Figure 8. Layout methods to organize story slices according to di ff erent perspectives: (a) time cycles, (b) locations, (c) users, and (d) keywords. 6
Storytelling Presenting results in an accessible way! Figure 8. Layout methods to organize story slices according to di ff erent perspectives: (a) time cycles, (b) locations, (c) users, and (d) keywords. 7
Storytelling ‣ Helps non-experts connect the dots ‣ E ff ectively communicates insights ‣ Provides meaning and context to data 8
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 9
The Gap 10
The Gap 11
The Gap 12
The Gap How do we move from analysis to storytelling? 13
The Gap How do we move from analysis to storytelling? 14
The Gap Multiple approaches but what we need is a data analysis to storytelling pipeline How do we move from analysis to storytelling? 15
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 16
Past Work ‣ Need for organizing story contents ‣ Existing works compose stories with annotated screen shots of analytic displays ‣ Provenance systems rely on completed analysis bookmarks ‣ Other systems rely on completed visualizations 17
Past Work Inherits visual complexity of ‣ Need for organizing story contents analytic displays ‣ Existing works compose stories with annotated screen shots of analytic displays ‣ Provenance systems rely on completed analysis bookmarks ‣ Other systems rely on completed visualizations 18
Past Work Inherits visual complexity of ‣ Need for organizing story contents analytic displays ‣ Existing works compose stories with annotated screen shots of analytic displays No current systems are integrated into ‣ Provenance systems rely on completed analysis the analysis bookmarks workflow! ‣ Other systems rely on completed visualizations 19
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 20
Bridging the Gap Figure 6. Proposed framework for bridging the gap. 21
Bridging the Gap E xtend analysis and storytelling workflows toward each other Figure 6. Proposed framework for bridging the gap. 22
Bridging the Gap E xtend analysis and storytelling Story slice workflows toward generation each other Figure 6. Proposed framework for bridging the gap. 23
Bridging the Gap E xtend analysis and storytelling Story slice workflows toward generation each other C ontent preparation Figure 6. Proposed framework for bridging the gap. 24
Story Slices ‣ Structured representations of findings during the course of analysis ‣ Define what facets (information) to capture 25
Story Slices ‣ Structured representations of findings during the course of analysis ‣ Define what facets (information) to capture VAST Challenge, analyze the circumstances of an epidemic outbreak via microblogs 26
Story Slices ‣ Structured representations of findings during the course of analysis ‣ Define what facets (information) to capture Epidemic and city were fictitious don’t worry VAST Challenge, analyze the circumstances of an epidemic outbreak via microblogs 27
Analytic Phase Produce story slices during Analyze tasks Figure 5. Visual analytics phase. After topic modelling, analysts explore the collection of messages from four perspectives and extract story slices. 28
Story Slices … … 29
Story Slices … … Facets help guide visual arrangements in story contents 30
Story Slices … … Facets help guide visual arrangements Story slices are the in story contents input to the story synthesis phase 31
Facet Types ‣ Discrete categories with no relationships ‣ Discrete entities with domain-specific relationships ‣ Linearly ordered elements ‣ Linearly ordered elements with distances ‣ Time ‣ 1D/2D/3D space 32
Facet Types ‣ Discrete categories with no relationships ‣ Discrete entities with domain-specific relationships ‣ Linearly ordered elements ‣ Linearly ordered elements with distances ‣ Time ‣ 1D/2D/3D space Facets specify possible arrangements and aggregations 33
Facet Types ‣ Discrete categories with no relationships ‣ Discrete entities with domain-specific relationships ‣ Linearly ordered elements ‣ Linearly ordered elements with distances ‣ Time These are very similar to the Abstract Data ‣ 1D/2D/3D space Facets specify possible Types in VAD… arrangements and aggregations 34
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 35
Story Synthesis Story synthesis creates story contents from story slices by… ‣ Aggregating and summarizing story slices according to facets/measurements ‣ Embedding details into views/layouts ‣ Arranging story slices onto layouts according to facets/ relationships ‣ Showing facets by priority and from multiple perspectives ‣ Annotating 36
Story Synthesis Story synthesis creates story contents from story slices by… ‣ Aggregating and summarizing story slices according to facets/measurements ‣ Embedding details into views/layouts ‣ Arranging story slices onto layouts according to facets/ relationships ‣ Showing facets by priority and from multiple perspectives Story synthesis does ‣ Annotating not include the design of the final story! 37
Story Synthesis Figure 11. Output from story synthesis phase performed by sociology researcher. 38
Story Synthesis story slice keys Figure 11. Output from story synthesis phase performed by sociology researcher. 39
Story Synthesis story slice attributes story slice keys Figure 11. Output from story synthesis phase performed by sociology researcher. 40
Story Synthesis Channels are used to convey facet information Figure 11. Output from story synthesis phase performed by sociology researcher. 41
New Framework 1. Define types and structures of story slices - What facets or patterns may be used from the data analysis phase? 2. Design representations of story slices - What data structure will be used to contain slice information? How will slices be visualized in the synthesis workspace? 3. Define story synthesis support functions . 4. Design the visual analytics system, including support for discovery of potential story slices. 42
New Framework 1. Define types and structures of story slices - What facets or patterns may be used from the data analysis phase? 2. Design representations of story slices - What data structure will be used to contain slice information? How will slices be visualized in the synthesis workspace? 3. Define story synthesis support functions . This framework can be used in designing visual 4. Design the visual analytics system, including support for analytics systems that discovery of potential story slices. provide support for story synthesis. 43
Background Research Problem Related Works Bridging the Gap Proposed Solution Thoughts 44
Good ☺ Need for framework is well backed by current research ☺ Story slices help reason about abstract data more ☺ Story slices capture model of analysis concept generation ☺ Evaluated by two sociology experts (positive feedback) 45
Bad ☹ Paper didn’t do a great job proving generality of framework ☹ Majority of examples used online text message data ☹ Paper structure could be improved to help understanding ☹ Both experts only interviewed for 1.5h using similar data ☹ Is this framework better suited for messaging data? 46
Recommend
More recommend
