Improving Engagement of Animated Visualization with Visual Foreshadowing
aa r X i v : . [ c s . H C ] S e p © 2020 IEEE. This is the author’s version of the article that has been published in the proceedings of IEEEVisualization conference. The final version of this record is available at: xx.xxxx/TVCG.201x.xxxxxxx/ Improving Engagement of Animated Visualization withVisual Foreshadowing
Wenchao Li * The Hong Kong University ofScience and Technology
Yun Wang † Microsoft Research Asia
Haidong Zhang ‡ Microsoft Research Asia
Huamin Qu § The Hong Kong University ofScience and Technology A BSTRACT
Animated visualization is becoming increasingly popular as a com-pelling way to illustrate changes in time series data. However, main-taining the viewers focus throughout the entire animation is difficultbecause of its time-consuming nature. Viewers are likely to becomebored and distracted during the ever-changing animated visualiza-tion. Informed by the role of foreshadowing that builds the expec-tation in film and literature, we introduce visual foreshadowing toimprove the engagement of animated visualizations. In specific, wepropose designs of visual foreshadowing that engage the audiencewhile watching the animation. To demonstrate our approach, webuilt a proof-of-concept animated visualization authoring tool thatincorporates visual foreshadowing techniques with various styles.Our user study indicates the effectiveness of our foreshadowingtechniques on improving engagement for animated visualization.
Index Terms:
Human-centered computing—Visualization—Visualization design and evaluation methods; Computingmethodologies—Computer graphics—Animation
NTRODUCTION
In recent years, animated visualization has been widely adopted toshow changes in time series data, because of its inherent nature ofpresenting temporal evolution over time [11, 25]. When used ap-propriately, animated visualization can be attractive and effectiveto convey information [18, 29]. A well-known example is HansRosling’s Gapminder presentation using the animated bubble chartto review the financial and physical well-being changes of differentcountries over 200 years [26]. This work has garnered more than8 million views from all over the world [21] and has drawn consid-erable public attention toward data visualization. Since then, moreand more animated visualizations have emerged to present changesin data. For example,
Top 15 Best Global Brands Ranking [27] and
The History of the World’s Best Go Players [2] use an animated barchart to show changes in rankings.However, without additional guidance from a presenter, generalaudience might feel difficult to pay attention to key informationand reluctant to consume an entire animation. Although animatedvisualizations are visually appealing, people are not sure where tofocus when multiple changes simultaneously occur. Moreover, theever-changing nature of animated visualization makes it hard forpeople to retain their attention for a long time.In film and literature, foreshadowing is widely adopted as an im-portant narrative tactic to engage viewers and readers [8]. This nar-rative tactic uses implications early in a story to indicate the subse-quent emergence of a relevant occurrence in the plot [24,32]. There-fore, we propose to adopt visual foreshadowing for data storytelling * e-mail: [email protected] † e-mail: [email protected] ‡ e-mail: [email protected] § e-mail: [email protected] to improve the engagement of animated visualization. Visual fore-shadowing guides audience’s attention to significant changes in dataand raise expectations for forthcoming events.In this study, we formulate and apply visual foreshadowing onanimated bar chart, which is a common animated visualization forranking changes. We demonstrate four examples of visual fore-shadowing that can be categorized into two major types: explicit foreshadowing (that openly suggests the outcome) and implicit fore-shadowing (that leaves subtle clues by hinting the relevant items).To learn the efficacy of engagement enhancement of our approach,we implement a proof-of-concept system that generates animatedbar charts with diverse visual foreshadowing techniques. With thisnew animated visualization authoring tool, we support the easy cre-ation of visual foreshadowing effects for specific item(s). Our sys-tem facilitates adding, previewing, and editing visual foreshadow-ing for the selected items with different temporal ranking data. Ouruser study results suggest that the proposed visual foreshadowingtechniques are useful to engage the audience of animated visualiza-tion. ELATED W ORK
A large variety of animations being applied to different domains.Thus, the need for emphasizing crucial parts of animation has moti-vated researchers to study attention guidance for dynamic visualiza-tion [10,15,22,31]. For example, in the education domain, De Kon-ing et al. [13] attempted to transfer successful cueing approachesfrom static visualization to animation for instructional design. Theauthors proposed a framework that classifies three cueing functions,selection, organization, and integration. This work suggested devel-oping visual cues for animation instead of borrowing the effectiveones for static representations. The more recent work of De Kon-ing et al. [12], summarized visual cues to direct learner attention tokey information, including basic arrows, colored circles, and othervisual cues embedded within graphical entities of dynamic visual-ization. In an effort to locate task-relevant information, Etemadpouret al. [16] and Chen et al. [9] applied motions to data points, andDe Koning et al. [14] studied the role of presentation speed in atten-tion cueing. In dynamic narrative visualization, Waldner et al. [30]studied the flicker effect and found a good trade-off between attrac-tion effectiveness and the subjective annoyance caused by flicker-ing. Many of the considerations from these works informed thedesign of our visual foreshadowing.Despite the availability of existing studies, the topic of how vi-sual cues help improve engagement in animation or dynamic visual-ization is still under exploration. The work by Wang et al. [32] is themost relevant to our approach. The authors first adopted the term“foreshadowing” to help enhance the narratives of clickstream datavisualization. However, their techniques are employed in the time-line, and the visual effects are limited to animated stacked graph. Inthe current research, we formally define visual foreshadowing andpropose design examples for animated visualization.
ISUAL F ORESHADOWING
Foreshadowing is a narrative tactic that is widely used in film andliterature. It often appears in the early stages of a plot because it can1 subtlety create tension and set viewers’ expectations regarding howthe story will unfold. We conduct an extensive literature reviewof the foreshadowing techniques used in the literature, film, andvideo games domains [1,4,6,20,23,28], analyze the existing visualcues in data visualization [19, 30, 32], and formally define visualforeshadowing. Taking bar chart race as an example, we initializevisual foreshadowing designs for animated visualization.
We see visual foreshadowing as a new form of animation effect thatappears before the critical events during the playback of animatedvisualization to set the viewer’s expectations. Basing from previousresearch on animation [17], we define visual foreshadowing for an-imated visualization from three perspectives, namely, visual effect , timing , and duration . Accordingly, visual foreshadowing can beformalized as a 3-tuples: Visual Foreshadowing : = ( visual effect(s) , timing(s) , duration (s) ) . Visual foreshadowing can have one or multiple visual effects .The visual effects in visual foreshadowing can be a textual state-ment or the ones that are attached to specific visual elements. Un-der the framework, different visual cues (e.g., flickering or pointingwith arrows) can be easily introduced to extend the foreshadowingeffects.Foreshadowing can be achieved implicitly or explicitly [5, 32].Following this taxonomy, the visual effects adopted in visual fore-shadowing should also have two categories.
Explicit foreshadow-ing occurs when an outcome is directly indicated, which explicitlygives viewers a piece of information to entice them to want more.Therefore, the audience is directly informed of the key informationor the final outcome.
Implicit foreshadowing appears when an out-come is indirectly hinted, which leaves subtle clues about a futureevent by suggesting related items. The upcoming event is only ap-parent to viewers after it has occurred. On the basis of prior workon visual cue preference and highlighting interventions for staticvisualizations [7], the visual effects for indicating relevant visual el-ements in indirect foreshadowing can include adding a contour andsetting the transparency. By using the visual cues, the author candraw viewers’ attention to the target items.Given the specified visual effect, timing is used to specify whenthe visual effects start, while duration is to control the time lengthof the visual effects. Timing and duration should be chosen foreach foreshadowing visual effect. The author can define when tostart building the viewer’s expectations and when to stop before thecrucial event occurs.
The design of foreshadowing can be numerous. Considering thedesign space of possible visual cues [19], we strategically selectthe common cueing methods to ease the burden of understandingthe visual designs. We demonstrate two explicit and two implicitforeshadowing techniques to enhance animated visualization withtwo different datasets.•
Prologue.
Prologue is an explicit foreshadowing example. In-spired by the study of Kong et al. [20] that the perceived mainmessage of a visualization can be influenced by the slant ofthe title, the foreshadowing design is to insert additional textdescription to suggest the forthcoming events (see Fig. 1(a)).After noticing the caption, the audience’s focus is guided tothe target bar so that they can see how the bar animates tomeet their expectation.•
Pre-scene.
Pre-scene is an explicit foreshadowing example.The goal of the design is to arouse the viewer’s interest to seehow the selected bar would change to the final state. Thisdirect foreshadowing design builds the expectation visually,where the final ranking and the length of the bar are directly (cid:11)(cid:68)(cid:12)(cid:3)(cid:51)(cid:85)(cid:82)(cid:79)(cid:82)(cid:74)(cid:88)(cid:72)(cid:3)(cid:11)(cid:69)(cid:12)(cid:3)(cid:51)(cid:85)(cid:72)(cid:16)(cid:86)(cid:70)(cid:72)(cid:81)(cid:72)(cid:3)
Figure 1: The visual effects of two explicit foreshadowing be-fore the key event occurs. The Prologue effect is designed tobuild the expectation literally, while the Pre-scene effect is de-signed to achieve it visually. Data are collected from Spotify(https://spotifycharts.com/regional), which provides the historicalpopularity rankings of the songs streamed by the users. The colorsof the bars encode the type of the singers. visualized. As illustrated in Fig. 1(b), the final state of the iSpy item is directly shown in the animated visualization. Peoplemay be curious about how the suggested item of the currentstate turns out.•
Contour.
Contour is an implicit foreshadowing example. Thevisual effect of the foreshadowing technique is drawing a con-tour around the target bar to attract the audience’s attentionbefore important change occurs. In contrast to the explicitforeshadowing techniques, this type of visual foreshadowingonly highlights the relevant items without disclosing what willhappen in the future. As shown in Fig. 2(a), the bar with con-tour is suggesting some interesting change will happen to it.•
De-emphasis.
De-emphasis is an implicit foreshadowingexample. Similar to the Contour effect, interesting itemswill be highlighted before key events occur. As shown inFig. 2(b), the De-emphasis effect is implemented by settingtransparency. The relevant items hold the original trans-parency, whereas the transparency of those irrelevant bars isset to 20%.
ISUAL F ORESHADOWING A UTHORING
To evaluate the visual foreshadowing design, we built a proof-of-concept system for animated visualization authoring. The user inter-face of the prototyping system consists of a foreshadowing configu-ration panel (Fig. 3(a)), a basic animation setting widget (Fig. 3(b)),an animation preview panel (Fig. 3(c)), a timeline panel (Fig. 3(d)),and a data loading panel (Fig. 3(e)). We provide additional details2 (cid:11)(cid:68)(cid:12)(cid:3)(cid:38)(cid:82)(cid:81)(cid:87)(cid:82)(cid:88)(cid:85)(cid:3)(cid:11)(cid:69)(cid:12)(cid:3)(cid:39)(cid:72)(cid:16)(cid:72)(cid:80)(cid:83)(cid:75)(cid:86)(cid:76)(cid:86)(cid:3) about the system with an example scenario for the creation of ananimated bar chart.Before applying the visual foreshadowing to the animated barcharts, the user needs to load the temporal data of ranking changes.The underlying data can be updated by editing the data table. Asshown in Fig. 3(e), each row of the data table represents a rankingitem, while the columns show the temporal dimension.After loading the CSV data file of the chart, an initial staticbar chart will be displayed in the preview panel (Fig. 3(c)), andthe animated visualization illustrating the ranking changes can beplayed. To apply the visual foreshadowing to the animated barchart, the user need to select the specific items and create the as-sociated visual foreshadowing effect in the foreshadowing editingwidget (Fig. 3(a)). In our current proof-of-concept implementation,we provide the control of visual foreshadowing in terms of fore-shadowing type, visual effects, and time range.For instance, a user wants to add an explicit foreshadowing effectto suggest the rank of the Coca-Cola company in 2019. The usercan obtain the corresponding bar by choosing the item in the itemlist. Then, the user needs to specify the settings for explicit fore-shadowing by selecting the effect type, providing relevant words,and choosing the time range to show the visual effects. Finally,the resulting foreshadowing effect with time range is indicated inthe timeline widget (Fig. 3(d)). The user can preview the synthe-sized animated visualization with additional visual foreshadowingeffects.
VALUATION
Our approach is designed to build the expectation and guide the au-dience’s attention to watch the animated visualization in an engag-ing manner. Our preliminary observations show that the viewersare often unaware where to focus without a narrator presenting, ex-plaining, and highlighting the key messages when frequent changesoccur in the ranking data. Through introducing timely indicators,we provide additional clues for the animated visualization of tem-poral data. To understand if our visual foreshadowing techniquescould enhance the engagement of animated visualization, we con-ducted a user study to compare the effects of playing the rankinganimation with and without visual foreshadowing effects.
We recruited 12 participants (5 males and 7 females; aged 20-27[median = 23.3, SD = 2.5]), denoted as P1-P12, for this study. Theparticipants were mostly graduate and undergraduate students ina local university and with different backgrounds (i.e., computerscience, life science, and arts). We started our user study by in-troducing the dataset background and the encoding scheme of theproposed visualization. We used two different ranking change vi-sualizations of global top brands with two different time periodsand provided the participants with two visual foreshadowing set-tings (i.e., with and without) for comparison. Different types ofvisual foreshadowing designs are covered under the condition withforeshadowing. Then, we asked each participant to complete a sur-vey about the overall user engagement of the animated visualiza-tion. The survey consisted of 12 subjective questions referencedfrom previous studies [3]. Each question was rated on a 7-pointLikert scale (1 = strongly disagree, 7 = strongly agree). In addi-tion, we introduced the proof-of-concept system to each participantand encouraged them to freely explore the system to generate visualforeshadowing for the ranking animations. Finally, the user studyended with a semi-structured interview for collecting feedback fromeach participant. Each participant took approximately 20 minutesto complete the whole study, including the informal interview.
Table 1 shows the results of three categories of engagement mea-sures. Overall, the animated visualization with visual foreshadow-ing achieves higher subjective ratings for the one without visualforeshadowing, indicating the effectiveness of our visual foreshad-owing design on engagement improvement. In the interview ses-sion, almost all the participants valued the visual foreshadowingtechniques. One participant (P3) commented that “
The additionalvisual effects are useful and make the animation more like a story.Otherwise, I don’t know where to look at and forget almost all thechanges. ” Another participant (P12) said “
I built the anticipationwhen I saw the upcoming icons on the timeline, which sets me aclearer goal for where to concentrate on. ”However, some participants also offered suggestions for the fore-shadowing design. One of the participants (P10) said, “
I think itwould be very interesting to include the slow-down and zoomingeffects to draw our attentions. ” A participant (P8) pointed out thata limited number of visual foreshadowing can improve the overallengagement of the animation. However, adding excessive visualforeshadowing would be overwhelming to the audience. Moreover,when visual foreshadowing is introduced, the audience expected itto be placed long enough before the event occurs. This setting willgive them more joy when they come back through the data storyand find the message left before.
ISCUSSION
Future Extensions of Visual Foreshadowing Designs
Thegoal of the visual foreshadowing in our work is to entice the viewerto consume the entire animation instead of the one with tedious3 a b c d e Figure 3: Our prototype system for visual foreshadowing authoring. The system mainly consists of (a) editing widgets for visual foreshadowing,(b) basic settings for animated visualization, (c) animated visualization preview, (d) timeline for visual foreshadowing, and (e) data view. Theauthor can add visual foreshadowing for a specific item and get an overall review of the timeline.Table 1: Average ratings of the user engagement survey questions(1 = strongly disagree, 7 = strongly agree).
Assessment Foreshadowing Mean SD
Without 4.13 0.61Enjoyment With 6.54 0.54Without 3.71 1.03Focused Attention With 6.17 0.72Without 4.50 1.04Cognitive Involvement With 6.04 0.58changes. As a first step, we propose four visual foreshadowing de-sign examples focus on the animated bar chart of temporal rank-ing data. However, visual foreshadowing should not be limited tospecific charts types. For example, more visual effects and evendynamic visual cues can be introduced and combined with the tem-poral aspect of our visual foreshadowing. Moreover, other formsof animated visualizations can be adopted. For example, the fore-shadowing techniques of direct text description and indirect itemindication can be applied to animated line charts or animated scat-terplots without too much modification. The visual effects, giventheir timing and duration settings, can still be applied to these com-mon animated visualizations.
In-depth Investigation on Effectiveness
In our user study,we investigated visual foreshadowing with several visual effectsgenerally. Interestingly, participants had a stronger preference onthe foreshadowing designs that only leaves subtle clues rather thanthe ones that openly suggest the events was empirically observed.For instance, compared with participants that tested with the Pre-scene effect, participants that tested with the De-emphasis effectwere more willing to wait until the end of an animated transitionto obtain the final outcome. However, which visual foreshadowing design and which foreshadowing setting (e.g., timing, and duration)are the most effective ones remain unclear. What about the combi-nations of visual foreshadowing for engagement improvement? Inaddition, we studied general engagement improvement on a rathersmall tested data. Whether or not viewers will behave differently inmore complicated data is also interesting to study. Examining theseideas requires further investigation, which is outside the scope ofthis work. We encourage follow-up studies to explore the differentaspects of visual foreshadowing used in animated visualization.
ONCLUSION
In this work, we introduced visual foreshadowing, a design thatenhances animated visualization with effective visual cues aheadof time. The visual cues with proper timing prepare the audiencewith critical patterns of temporal data and enhance temporal changerecognition. We build a proof-of-concept system to support fore-shadowing authoring and further investigation. We demonstrateour visual foreshadowing approach with a qualitative evaluation.Results show that our method can guide participants’ preparationfor the upcoming events in the generated animated ranking visu-alization. The subjective preference ratings reveal that animatedvisualization with visual foreshadowing are engaging to show tem-poral data changes. In the future, we plan to investigate other de-sign choices (e.g., the design of visual cues) with different levels ofdata complexities. We will also integrate visual foreshadowing intoother forms of animated visualization, such as dynamic line chartand scatter plot, to provide effective animated narrative visualiza-tions. A CKNOWLEDGMENTS
The authors thank the anonymous reviewers for their valuable com-ments. This work was supported in part by a grant from Microsoft4
Research Asia. R EFERENCES [1] E. Aarseth. A narrative theory of games. In M. S. El-Nasr, M. Con-salvo, and S. K. Feiner, eds.,
International Conference on the Founda-tions of Digital Games, FDG ’12, Raleigh, NC, USA, May 29 - June01, 2012 , pp. 129–133. ACM, 2012.[2] Abacaba. The history of the world’s best go players. Available at .[3] F. Amini, N. H. Riche, B. Lee, J. Leboe-McGowan, and P. Irani.Hooked on data videos: assessing the effect of animation and pic-tographs on viewer engagement. In
Proceedings of the 2018 Inter-national Conference on Advanced Visual Interfaces, AVI 2018, Cas-tiglione della Pescaia, Italy, May 29 - June 01, 2018 , pp. 21:1–21:9,2018. doi: 10.1145/3206505.3206552[4] B. Bae, Y. Cheong, and D. Vella. Modeling foreshadowing in narra-tive comprehension for sentimental readers. In H. Koenitz, T. I. Sezen,G. Ferri, M. Haahr, D. Sezen, and G. Catak, eds.,
Interactive Story-telling - 6th International Conference, ICIDS 2013, Istanbul, Turkey,November 6-9, 2013, Proceedings , vol. 8230 of
Lecture Notes in Com-puter Science , pp. 1–12. Springer, 2013.[5] B.-C. Bae and R. M. Young. A use of flashback and foreshadowingfor surprise arousal in narrative using a plan-based approach. In
Jointinternational conference on interactive digital storytelling , pp. 156–167. Springer, 2008.[6] D. Bordwell.
Narration in the Fiction Film . University of WisconsinPress, 1985.[7] G. Carenini, C. Conati, E. Hoque, B. Steichen, D. Toker, and J. T.Enns. Highlighting interventions and user differences: informingadaptive information visualization support. In
CHI Conference on Hu-man Factors in Computing Systems, CHI’14, Toronto, ON, Canada -April 26 - May 01, 2014 , pp. 1835–1844, 2014.[8] S. B. Chatman.
Story and discourse: Narrative structure in fiction andfilm . Cornell University Press, 1980.[9] H. Chen, S. Engle, A. Joshi, E. D. Ragan, B. F. Yuksel, and L. Har-rison. Using animation to alleviate overdraw in multiclass scatterplotmatrices. In R. L. Mandryk, M. Hancock, M. Perry, and A. L. Cox,eds.,
Proceedings of the 2018 CHI Conference on Human Factors inComputing Systems, CHI 2018, Montreal, QC, Canada, April 21-26,2018 , p. 417. ACM, 2018.[10] F. Chevalier, P. Dragicevic, and S. Franconeri. The not-so-staggeringeffect of staggered animated transitions on visual tracking.
IEEETrans. Vis. Comput. Graph. , 20(12):2241–2250, 2014.[11] F. Chevalier, N. H. Riche, C. Plaisant, A. Chalbi, and C. Hurter. Ani-mations 25 years later: New roles and opportunities. In
Proceedings ofthe International Working Conference on Advanced Visual Interfaces,AVI 2016, Bari, Italy, June 7-10, 2016 , pp. 280–287, 2016.[12] B. B. de Koning and H. Jarodzka. Attention guidance strategies forsupporting learning from dynamic visualizations. In
Learning fromdynamic visualization , pp. 255–278. Springer, 2017.[13] B. B. De Koning, H. K. Tabbers, R. M. Rikers, and F. Paas. To-wards a framework for attention cueing in instructional animations:Guidelines for research and design.
Educational Psychology Review ,21(2):113–140, 2009. [14] B. B. de Koning, H. K. Tabbers, R. M. J. P. Rikers, and F. Paas. At-tention cueing in an instructional animation: The role of presentationspeed.
Computers in Human Behavior , 27(1):41–45, 2011.[15] F. Du, N. Cao, J. Zhao, and Y. Lin. Trajectory bundling for animatedtransitions. In
Proceedings of the 33rd Annual ACM Conference onHuman Factors in Computing Systems, CHI 2015, Seoul, Republic ofKorea, April 18-23, 2015 , pp. 289–298, 2015.[16] R. Etemadpour and A. G. Forbes. Density-based motion.
InformationVisualization , 16(1):3–20, 2017.[17] T. Ge, Y. Zhao, B. Lee, D. Ren, B. Chen, and Y. Wang. Canis: Ahigh-level language for data-driven chart animations.
Comput. Graph.Forum , 39(3):607–617, 2020.[18] J. Heer and G. G. Robertson. Animated transitions in statistical datagraphics.
IEEE Trans. Vis. Comput. Graph. , 13(6):1240–1247, 2007.[19] H. K. Kong, Z. Liu, and K. Karahalios. Internal and external visualcue preferences for visualizations in presentations.
Comput. Graph.Forum , 36(3):515–525, 2017.[20] H. K. Kong, Z. Liu, and K. Karahalios. Frames and slants in titles ofvisualizations on controversial topics. In R. L. Mandryk, M. Hancock,M. Perry, and A. L. Cox, eds.,
Proceedings of the 2018 CHI Confer-ence on Human Factors in Computing Systems, CHI 2018, Montreal,QC, Canada, April 21-26, 2018 , p. 438. ACM, 2018.[21] R. Kosara and J. D. Mackinlay. Storytelling: The next step for visual-ization.
IEEE Computer , 46(5):44–50, 2013.[22] J. Lu, J. Wang, H. Ye, Y. Gu, Z. Ding, M. Xu, and W. Chen. Illustratingchanges in time-series data with data video.
IEEE Computer Graphicsand Applications , 40(2):18–31, 2020.[23] P. Martin.
The Writer’s Guide to Fantasy Literature: From Dragon’sLair to Hero’s Quest : how to Write Fantasy Stories of Lasting Value .Writer Books, 2002.[24] G. Moura. What is foreshadowing? Available at .[25] G. G. Robertson, R. Fernandez, D. Fisher, B. Lee, and J. T. Stasko.Effectiveness of animation in trend visualization.
IEEE Trans. Vis.Comput. Graph. , 14(6):1325–1332, 2008.[26] H. Rosling. Gapminder. Available at .[27] TheRankings. Top 15 best global brands ranking. Available at .[28] S. Turner.
The Creative Process: A Computer Model of Storytellingand Creativity . L. Erlbaum, 1994.[29] B. Tversky, J. B. Morrison, and M. B´etrancourt. Animation: can itfacilitate?
Int. J. Hum.-Comput. Stud. , 57(4):247–262, 2002.[30] M. Waldner, M. L. Muzic, M. Bernhard, W. Purgathofer, and I. Viola.Attractive flicker - guiding attention in dynamic narrative visualiza-tions.
IEEE Trans. Vis. Comput. Graph. , 20(12):2456–2465, 2014.[31] Y. Wang, D. Archambault, C. E. Scheidegger, and H. Qu. A vec-tor field design approach to animated transitions.
IEEE Trans. Vis.Comput. Graph. , 24(9):2487–2500, 2018. doi: 10.1109/TVCG.2017.2750689[32] Y. Wang, Z. Chen, Q. Li, X. Ma, Q. Luo, and H. Qu. Animated nar-rative visualization for video clickstream data. In
SIGGRAPH ASIA2016, Macao, December 5-8, 2016 - Symposium on Visualization , pp.11:1–11:8, 2016., pp.11:1–11:8, 2016.