Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays Mohammad Chegini∗ , Lin Shao∗ , Dirk J. Lehmann† , Keith Andrews‡ and Tobias Schreck∗ ∗ Institute of Computer Graphics and Knowledge Visualisation, Graz University of Technology, Austria Email: {m.chegini, l.shao, t.schreck}@cgv.tugraz.at † University of Magdeburg, Germany Email: dirk@isg.cs.uni-magdeburg.de ‡ Institute of Interactive Systems and Data Science, Graz University of Technology, Austria Email: kandrews@tugraz.at Abstract—Large vertically-mounted high-resolution multi- Although there are studies about collaborative interaction with touch displays are becoming increasingly available for interactive large displays (e.g. [7], [8]), they usually focus on single- data visualisation. Such devices are well-suited to small-team user interaction [9]. Since typical multi-touch interactions do collaborative visual analysis. In particular, the visual analysis of large high-dimensional datasets can benefit from high-resolution not support collaboration, more research needs to be done on displays capable of showing multiple coordinated views. cooperative gestures, modalities and the dynamics of group This paper identifies some of the advantages of using large, work around these devices. Cooperative gestures are known to high-resolution displays for visual analytics in general, and enhance the sense of teamwork and increase the participation introduces a set of interactions to explore high-dimensional of team members [10]. datasets on large vertically-mounted high-resolution multi-touch displays using scatterplots. A set of touch interactions for col- Screen size and resolution are particularly important for laborative visual analysis of scatterplots have been implemented information visualisation of multivariate datasets. Having a and are presented. Finally, three perception-based level of detail large display allows multiple, linked views, such as scatter- techniques are introduced for such displays as a concept for plot matrices and parallel coordinates [11] to be provided further implementation. simultaneously. If the screen is not high-resolution, the user experience of near distance interaction decreases significantly. I. I NTRODUCTION For instance, on screens with less than sixty pixels per inch, Large high-resolution displays are becoming an affordable the user is not able to read from the screen up-close [12]. Fur- option for the visualisation of data [1]. Large displays have thermore, users can make more observations with less effort proved to be effective for tasks such as comparative genomics using physical navigation (e.g., walking) rather than virtual analysis [2], graph topology exploration [3], and sensemaking [1]. More screen space can be used to either provide a better [4]. Large vertically-mounted (landscape-orientation) high- overview of a dataset or to provide more details of a portion resolution multi-touch displays are particularly effective for of it. For example, users can see both an entire scatterplot collaborative analysis by small teams. However, previous matrix, specific scatterplots, and parallel coordinates plots at research has often focused on horizontally-mounted tabletop the same time. As a result, users may have the opportunity to surfaces or vertically-mounted displays with more distant gain more insight into large datasets. interaction [5]. In this paper, a set of user interactions to Previous studies [5] suggest that vertically-mounted displays support scatterplot matrices analysis on vertically-mounted are more suited to parallel tasks within a group, due to reduced displays are introduced. These techniques help analysts to visual distraction and the possibility to share information efficiently select a scatterplot from scatterplot matrices and through physical navigation like turning the head or walking. explore it collaboratively. On tabletop displays, if users are not on the same side of the Some physical and virtual interactions with large displays table, the shared view often needs to be reoriented. were described in the previous literature. Modalities range This paper addresses the design gap between standard inter- from natural interactions like speech, body tracking, gaze, and action techniques for large, multi-touch displays and advanced gestures to the use of secondary control devices like mobile interaction techniques and visual feedback for collaborative phones, tablets, or Wii controllers [6]. Of these, multi-touch scatterplot and scatterplot matrix analysis. Design concepts interactions provide a fluid and intuitive interface suitable for for such interaction techniques have been implemented as a up-close interaction in front of the display by small groups. proof of concept and are presented. The techniques include 90 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays densely or sparsely distributed targets. They concluded that since the whole dataset fits on a larger display, sparse targets can be found faster. Multiple linked views are often used to gain a better understanding of a high-dimensional dataset. Such views are usually connected by techniques such as brushing or combined navigation [18]. Every view occupies space on display. If more space is available, additional views can be shown simultane- ously. Allowing the user to access multiple windows increases performance and satisfaction [19]. Isenberg et al. [20] present hybrid-image visualisation for data analysis, where two images are blended to achieve distance-dependent perception. This concept might be especially helpful for collaborative visual analysis tasks on vertically-mounted displays, where users Fig. 1. Two users collaboratively analyse a dataset on a large vertically- observe data from various distances. mounted multi-touch screen. User A on the left drags a Regression Lens, while user B on the right adapts the degree of the regression model using the floating toolbox. The display is an Eyevis 84-inch 4K/Ultra-HD 60Hz multi-touch LCD monitor with a resolution of 3840 × 2160. B. Visual Data Analysis and Multi-Touch Interaction Previous researchers proposed various interaction tech- scatterplot selection from scatterplot matrices, collaborative niques for large displays and multi-dimensional dataset inter- regression model analysis, and an extension of the Regression action on multi-touch displays. Ardito et al. [18] proposed Lens [13] to include a floating toolbox. As a proof of concept, a classification of large display interaction having five di- the techniques are developed on a large display. mensions: visualisation technology, display setup, interaction The paper is structured as follows: Section II discusses re- modality, application purpose, and location. Khan presented a lated work. Several novel interaction designs for collaborative survey of interaction techniques and devices for large, high- visual analysis of scatterplots on large displays are introduced resolution displays [6]. The survey categorises modalities of in Section III. The use case and current implementation of the interaction into speech, tracking, gestures, mobile phones, hap- proposed interaction techniques are described in Section IV. tic and other technologies such as gaze and facial expression. Section VI introduces the concept of perception-based level of Tsandilas et al. presented SketchSliders [21], a tool that visual detail. The paper concludes with a discussion of open provides a mobile sketching interface to create sliders which problems and future work in Section VII. interact with multi-dimensional datasets on a wall display. In II. R ELATED W ORK comparison, in this paper, interaction is performed directly At a high level, information visualisation systems consist of on the display rather than using a secondary touch device. two components: visual representation and interaction. Visual Zhai et al. [22] introduced gesture interaction for wall displays representation concerns the mapping from data to display [14]. based on the distance of the user from the screen. The gestures The interaction starts with a user’s intent to perform a task, can be performed in far or near mode. Unlike the techniques followed by a user action. The system then reacts and feedback described in this paper, the proposed interaction gestures are is given to the user [15]. It is essential to consider both visual not directly related to visual analytics tasks. Heilig et al. [23] representation and interaction when designing an application developed multi-touch scatterplot visualisation on a tabletop for information visualisation. display. Sadana and Stasko [24] proposed advanced techniques for scatterplot data selection on smaller touch-based devices, A. Visualisation on Large Displays such as tablets and smartphones, whereas this paper focuses Researchers in various fields are increasingly confronted on large multi-touch displays. with the challenge of visualising and exploring high- MultiLens supports various gestures for fluid multi-touch dimensional datasets [13], [16]. Keim argues that although exploration of graphs [25]. The Regression Lens [13] allows many traditional techniques exist to represent data, they are the user to interactively explore local areas of interest in often not scalable to high-dimensional datasets without suit- scatterplots by showing the best fitting regression models able analytical or interaction design [16]. inside the lens. The idea of visualising local regression models With the current size and resolution of typical computer is also studied by Matković et al. [26]. Rzeszotarski et al. [27] displays, it is challenging to represent entire datasets on one introduced Kinetica, a tool for exploring multivariate data by screen using techniques like scatterplot matrices or parallel physical interactions on multi-touch screens. Kister et al. [25] coordinates. The user is often forced to resort to panning and presented BodyLenses, a promising set of magic lenses for zooming, leading to frustration and longer task completion wall displays, which are mostly controlled by body interaction times. Ruddle et al. [17] conducted an experiment in which and therefore suitable for interacting with wall displays from participants searched maps on three different displays for a distance. 91 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays Fig. 2. On the left, a user is drags a Regression Lens with the right hand while Fig. 3. On the left, two users collaboratively analyse a scatterplot. Both users adjusting the lens with the left hand. On the right, a user drags a scatterplot create a regression model for a subset of selected data. The created models with the right hand while panning through the scatterplot matrix with the left are displayed in their partner’s respective lens as well, supporting comparison hand. of local data models. On the right, one user analyses a scatterplot, while their partner selects interesting plots in the scatterplot matrix and passes them over by holding the background and swiping the right hand. In comparison to this work, the aforementioned studies either focus on a different type of interaction and medium or are not designed for collaborative visual analytics tasks. terplots and scatterplot matrices on such devices. Some of the interaction techniques are based on the concept of the C. Collaborative Visualisation Regression Lens [13], which supports real-time regression Large displays are well-suited to collaboration [28], [29]. analysis of subsets of a scatterplot through lens selection Jakobsen and Hornbæk [5] conducted an exploratory study to and manipulation. With Regression Lens, a user can select understand group work with high-resolution multi-touch wall a local area in a scatterplot and observe the regression model displays. The study suggests that using this kind of display of selected points [13]. Shao et al. proposed operations to helps users to work more efficiently as a group and fluidly adjust and manipulate the regression model shown in the change between parallel and joint work. A large display ben- Regression Lens, such as changing the degree of the regression efits group working on a shared task, since users can operate model or inverting its axes. Figure 1 illustrates some of the on one common physical medium and share information on suggested collaborative gestures on an 84-inch 4K/ULTRA- it. HD@60HZ multi-touch LCD monitor produced by Eyevis Morris et al. [10] formalised the concept of cooperative [33]. The user on the left finds interesting scatterplots and gestures as a set of gestures performed by multiple users and passes them to the user on the right. The user on the right interpreted as a single task by the system. Liu et al. developed analyses the plots using the Regression Lens [13]. In the rest CoReach [9], a set of gestures for collaboration between two of this section, four interaction designs for both collaborative users over large multi-touch displays. Comparing the use of a and single scatterplot analysis are introduced. Later in Section large vertically-mounted display against two ordinary desktop IV, an implementation of these techniques is demonstrated. displays, Prouzeau et al. [30] concluded that groups obtain A. Lens and Floating Toolbox better results and communicate better on large, vertically- mounted displays. Magic lens techniques like DragMagics [34] and An experiment by Pedersen and Hornbæk [31] showed BodyLens [35] are used to explore local regions in a that users prefer horizontal surfaces over vertically-mounted visualisation. An extended version of the basic lens concept displays, but this result was limited to simple single-user provides for more fluid interaction with large multi-touch tasks and not collaborative tasks with different dynamics. displays. For instance, as shown in Figure 2, after a region of Vertically-mounted displays allow users to obtain an overview interest has been selected in a scatterplot using the dominant of their data by stepping back from the display and make it hand (here the right hand), a toolbox appears next to the possible to interact from afar as well as up close. Badam et other side of the lens (near the non-dominant hand), where al. [32] proposed a system for collaborative analysis on large the user can use sliders and touch buttons to adjust the lens. displays by controlling individual lenses through explicit mid- For example, the user can change the degree of the regression air gestures. model. The lens can be dragged with one hand, while being Although these studies are not directly related to collabo- adjusted with the second hand, thus potentially speeding up rative scatterplot analysis on large multi-touch displays, they performance. do provide valuable insights into the design process of such B. Two-Handed Interaction with Scatterplot Matrices systems. A scatterplot matrix consists of pairwise scatterplots ar- III. P ROPOSED I NTERACTION T ECHNIQUES ranged in a matrix, with dimensions typically labelled in the Current standard multi-touch interaction techniques are diagonal cells. Since the number of dimensions is usually not designed for collaboration on vertically-mounted high- high, panning and zooming within the scatterplot matrix is resolution displays [9]. Here, both single-user and collab- almost inevitable. With common multi-touch interactions, the orative interactions are proposed for the analysis of scat- scatterplot or dimension label is dragged to the corner of 92 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays Fig. 4. A user selects a scatterplot of interest from a scatterplot matrix by Fig. 5. A Regression Lens containing a cubic regression model is shown. At touching and holding the left hand on the scatterplot. Swiping with the right the left side of the Regression Lens, a floating toolbox with various options hand then passes the selected scatterplot to the right hand side of the display is visible. for more detailed analysis. of Figure 3, the user on the left side of the screen creates a the scatterplot matrix for panning. It is not feasible to zoom regression lens and regression model in blue. Meanwhile, the into or out of a scatterplot matrix while dragging another user on the right side of the screen creates their regression object. Based on two-handed interaction on tablets [36], a lens and regression model in red. Both users can see the other two-handed technique is proposed whereby the dominant hand user’s regression model reflected in their own regression lens. is responsible for dragging items, while the non-dominant hand performs common operations. As shown on the left side IV. I MPLEMENTATION of Figure 2, the user drags a scatterplot around to reorder Proof-of-concept interaction techniques for single-user and the plots in the scatterplot matrix. Panning is performed by collaborative analysis of scatterplots and scatterplot matrices the non-dominant hand. With this two-handed technique, the have been implemented on a vertically-mounted Eyevis 84- interactions needed to reorder scatterplots in a scatterplot inch multi-touch display with a resolution of 3840 × 2160 matrix can be reduced. pixels and a frame rate of 60 Hz. Figure 1 demonstrates a typical setup of the implemented application with two users C. Collaboration using Gestures working on the screen. On large vertically-mounted collaborative displays, it is not The prototype application is written in Java, using JavaFX always desirable to move from one side of the screen to for the user interface and the TUIO [37] and the TUIOFX the other to perform a task. Instead, collaborative gestures library [38] for multi-touch interaction. To enable multiple can be used to pass objects. Based on the ideas of Liu et users to work on the same screen with different widgets al. [9], collaborative gestures on scatterplots are proposed. and user interface elements at the same time, a concept In the right-hand side of Figure 3, the user on the left is called focusArea from the TUIOFX library is used [39]. The analyses a scatterplot, while the user on the right selects application follows the widely-used Model-View-Controller another scatterplot of interest. By holding the background of (MVC) architecture. the scatterplot matrix with one hand, and swiping with the other hand, the scatterplot is passed over to the partner. The V. U SE C ASE partner can then decide whether or not to load the scatterplot The use case for the prototype application is to improve for comparison. This technique can also be used for other interaction with the Regression Lens on multi-touch screens. tasks. For example, in Figure 4, the user selects a scatterplot The developed interaction techniques were tested with the of interest from a scatterplot matrix by touching and holding it well-known car dataset from the UCI Machine Learning with one hand (here, the left hand) and swipes the other hand Repository [40]. in the direction of the analysis panel to load that scatterplot For the interaction technique shown in Figure 1, user A (on for more detailed analysis. the left) and user B (on the right) select two different plots from the shared central area containing the scatterplot matrix. D. Collaborative Lens For this technique, the user holds and touches a scatterplot In collaborative analysis, visual feedback plays an essential with one hand and swipes to the right or left with the other role. When two analysts work on a vertically-mounted display hand to maximise it. This technique is elaborated in detail in without proper visual feedback, they need to communicate Section III-C. After that, users A and B select an area in the more and turn their heads more often. A collaborative lens scatterplot separately and toggle the Collaborative Lens option can help ameliorate this issue. As illustrated on the left side in the Floating Toolbox. As described in Section III-D, each 93 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays Fig. 6. The left and right panels are scatterplots for User A (left) and B (right) respectively. The central area of the screen contains a shared scatterplot matrix. User A on the left draws an arbitrary rectangle and is interested in the quadratic regression model of the selected records, shown in red. User B on the right chooses to observe the cubic regression model of the selected area, shown in blue. User A can see the cubic regression model of the right panel in dashed blue and user B can see the left panel regression model in dashed red. Selected scatterplots are highlighted in green in the scatterplot matrix. user is now able to observe the regression model of the other analysis. Three techniques are proposed to apply a perception- user in their regression lens. Figure 1 shows two users working based level of detail to scatterplots on large vertically-mounted side by side on a large vertically-mounted multi-touch display, high-resolution displays. after creating two separate Regression Lenses and toggling to Firstly, the concept of superpixels is similar to image the Double Lens option. The exact state of the screen is shown mosaics. A superpixel consists of a set of pixels in a small in Figure 6. A single Regression Lens with a floating toolbox rectangular area of the screen, for example a regular grid of is visible in Figure 5. say 50×50 pixels. The average colour, brightness, and contrast properties of superpixels can be used to visualise data for VI. P ERCEPTION - BASED L EVEL OF V ISUAL D ETAIL users farther from the screen. At the same time, the individual C ONCEPTS FOR S CATTERPLOTS colouring of pixels comprising a superpixel can be used to visualise more detailed information for users who are closer Users of large vertically-mounted high-resolution displays to the screen. may take up positions at varying distances from the display, Secondly, the concept of a Screen Progressive Visual Glyph and hence may perceive more or less detail in the display. (SPVG) utilises the colour, brightness, and contrast values of At greater distances from a large high-resolution display, less a glyph to encode different secondary information for closer detail is perceived. Here, perceived pixel density (PPD) is users. In Figure 8, the scatterplot on the left visually encodes defined as the number of pixels mapped to a single cell on two different classes (brown and cyan) in the data. This is the retina of the user’s eye. PPD increases quadratically as easily perceivable by a distant user. On the right, a user who is distance to the screen increases. The human perceptual system closer can make out an additional level of detail: the dots of the tends to average out too large PPD w.r.t. colour, brightness, scatterplot in fact contain an additional histogram representing and contrast [41], for example a red pixel and a green pixel the distribution of the related class in the data. In this case, is perceived as brown. the circles representing the mapped data points are SPVGs. The perceptual effect of averaging is well known, for The difference between SPVGs and superpixels is that SPVGs instance in the perception of secondary colours as a mixture encode different visual details of the same data at different of two primary colours or in the phenomena of metamerism. distances. In this way, they could be understood as a data More related effects include simultaneous contrast [42], after- filter concept as well. SPVGs can be placed on the screen images [43], and the Chubb effect [44]. Without delving too on demand and are not restricted to a regular grid, providing deeply into perception psychology, note that a sophisticated greater flexibility. theory for averaging effects are already available and well Thirdly, variational textures are related to halftone tech- described. For the purpose of this discussion with respect niques. Structural variations of an underlying texture can be to large high-resolution displays, it is sufficient to state that used to visually encode fine data details for users who are the effect of averaging a set of pixels is already exploited very close to the screen, while these details will immediately in practice by techniques such as image mosaics [41] and disappear when the user goes further away. halftone techniques [45], as illustrated in Figure 7. These proposed approaches for level of visual detail align Since PPD and related averaging effects are a function of well with Shneiderman’s mantra for information visualisation distance from the display, screen distance can be seen as an [46]: “Overview first, zoom and filter, details on demand”. In interactive parameter which can be exploited for visual data this case, distance from the screen is an additional degree of 94 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays Fig. 8. Screen Progressive Visual Glyphs (SPVGs): On the left, dots on a scatterplot representing items belonging to two classes (brown and cyan) are seen by distant users as simple dots. On the right, users who are closer to Fig. 7. On the left, a multi-image mosaic of the Mona Lisa [41]. On the the screen can perceive an additional histogram showing the distribution of right, an example of halftone dot sampling [45]. items. freedom, controlled by each user individually as they move support to recommend views for small collaborative team work closer to or further away from the display. The approaches on a large display. Moreover, adding group activity recognition are discussed as a concept and not implemented yet. and therefore pro-active interaction, can support collaboration by preventing information overload [48]. VII. D ISCUSSION AND F UTURE W ORK VIII. C ONCLUDING R EMARKS The concepts described in this paper are first designs of This paper presented challenges and solutions for collabora- appropriate touch interaction for the visual interactive analysis tive and single-task multi-touch interaction on large vertically- of scatterplot data on large vertically-mounted high-resolution mounted high-resolution displays. The techniques presented multi-touch displays. The interactions support small-group are well-suited for collaborative analysis tasks with scatterplots collaborative analysis, by exchanging patterns or settings from and scatterplot matrices. They are potentially generalisable for one user’s view to the others. The interaction design is other data exploration and visual analytics practices but require currently based on user selections, but is generalisable to further implementation and evaluation. Also, perception-based other basic techniques. The interaction techniques have been visualisation of scatterplots is introduced as a possible direc- implemented as a proof of concept. They still need to be tion for further research. evaluated with real users and real tasks as part of future work. Mapping out the design space for this combination R EFERENCES of visualisation and display device may well yield further [1] K. Reda, A. E. Johnson, M. E. Papka, and J. Leigh, “Effects of display interesting interaction designs. size and resolution on user behavior and insight acquisition in visual exploration,” in Proceedings of the 33rd Annual ACM Conference on The idea of exploiting perception-based level of detail for Human Factors in Computing Systems. ACM, 2015, pp. 2759–2768. the visualisation of scatterplots on large displays is new. [2] R. A. Ruddle, W. Fateen, D. Treanor, P. Sondergeld, and P. Ouirke, Detailed information can be rendered inside the marks of “Leveraging wall-sized high-resolution displays for comparative ge- nomics analyses of copy number variation,” in Biological Data Visu- the plot, becoming perceivable once users are closer to the alization (BioVis), 2013 IEEE Symposium on. IEEE, 2013, pp. 89–96. screen. Again, this is a proof of concept and requires further [3] A. Prouzeau, A. Bezerianos, and O. Chapuis, “Evaluating multi-user development and evaluation. selection for exploring graph topology on wall-displays,” IEEE Trans- actions on Visualization and Computer Graphics, 2016. While large high-resolution displays can improve the ex- [4] C. Andrews, A. Endert, and C. North, “Space to think: large high- ploration of large scatterplot spaces, further data analysis resolution displays for sensemaking,” in Proceedings of the SIGCHI support is needed to scale up with the number of data points conference on human factors in computing systems. ACM, 2010, pp. 55–64. and dimensions. Traditional techniques like cluster analysis [5] M. R. Jakobsen and K. Hornbæk, “Up close and personal: Collaborative and aggregation can help with scalability. Another relevant work on a high-resolution multitouch wall display,” ACM Transactions line of improvement is to adjust the view to the user’s on Computer-Human Interaction (TOCHI), vol. 21, no. 2, p. 11, 2014. [6] T. K. Khan, “A survey of interaction techniques and devices for large need and situation. In [47], the authors propose using eye high resolution displays,” in OASIcs-OpenAccess Series in Informatics, tracking to infer user interest and using this information to vol. 19. Schloss Dagstuhl-Leibniz-Zentrum fuer Informatik, 2011. recommend additional relevant but previously unseen views [7] K. Vogt, L. Bradel, C. Andrews, C. North, A. Endert, and D. Hutch- ings, “Co-located collaborative sensemaking on a large high-resolution for exploration. While that work was developed as a desktop display with multiple input devices,” Human-Computer Interaction– application, it might be interesting to incorporate eye-tracking INTERACT 2011, pp. 589–604, 2011. 95 Interaction Concepts for Collaborative Visual Analysis of Scatterplots on Large Vertically-Mounted High-Resolution Multi-Touch Displays [8] P. Isenberg, S. Carpendale, A. Bezerianos, N. Henry, and J.-D. Fekete, [29] P. Isenberg, T. Isenberg, T. Hesselmann, B. Lee, U. Von Zadow, and “Coconuttrix: Collaborative retrofitting for information visualization,” A. Tang, “Data visualization on interactive surfaces: A research agenda,” IEEE Computer Graphics and Applications, vol. 29, no. 5, pp. 44–57, IEEE Computer Graphics and Applications, vol. 33, no. 2, pp. 16–24, 2009. 2013. [9] C. Liu, O. Chapuis, M. Beaudouin-Lafon, and E. Lecolinet, “Coreach: [30] A. Prouzeau, A. Bezerianos, and O. Chapuis, “Trade-offs between a Cooperative gestures for data manipulation on wall-sized displays,” vertical shared display and two desktops in a collaborative path-finding in Proceedings of the 2017 CHI Conference on Human Factors in task,” in Proceedings of Graphics Interface 2017, 2017. Computing Systems. ACM, 2017, pp. 6730–6741. [31] E. W. Pedersen and K. Hornbæk, “An experimental comparison of touch [10] M. R. Morris, A. Huang, A. Paepcke, and T. Winograd, “Cooperative interaction on vertical and horizontal surfaces,” in Proceedings of the gestures: multi-user gestural interactions for co-located groupware,” in 7th Nordic Conference on Human-Computer Interaction: Making Sense Proceedings of the SIGCHI conference on Human Factors in computing Through Design. ACM, 2012, pp. 370–379. systems. ACM, 2006, pp. 1201–1210. [32] S. K. Badam, F. Amini, N. Elmqvist, and P. Irani, “Supporting visual [11] A. Inselberg, “The plane with parallel coordinates,” The visual computer, exploration for multiple users in large display environments,” in Visual vol. 1, no. 2, pp. 69–91, 1985. Analytics Science and Technology (VAST), 2016 IEEE Conference on. [12] M. Ashdown, P. Tuddenham, and P. Robinson, “High-Resolution Inter- IEEE, 2016, pp. 1–10. active Displays,” in Tabletops - Horizontal Interactive Displays, 2010, [33] “Eyevis display 84-inch,” 2017. [Online]. Available: http: pp. 71–100. //www.eyevis.de/en/products/lcd-solutions/4k-ultra-hd-lcd-monitors/ [13] L. Shao, A. Mahajan, T. Schreck, and D. J. Lehmann, “Interactive 84-inch-4k-uhd-lcd.html regression lens for exploring scatter plots,” in Computer Graphics [34] A. Prouzeau, A. Bezerianos, and O. Chapuis, “Towards road traffic Forum, vol. 36, no. 3. Wiley Online Library, 2017, pp. 157–166. management with forecasting on wall displays,” in Proceedings of the [14] J. S. Yi, Y. A. Kang, J. Stasko, and J. Jacko, “Toward a deeper 2016 ACM on Interactive Surfaces and Spaces. ACM, 2016, pp. 119– understanding of the role of interaction in information visualization.” 128. IEEE transactions on visualization and computer graphics, vol. 13, [35] U. Kister, P. Reipschläger, F. Matulic, and R. Dachselt, “Bodylenses: no. 6, pp. 1224–31, 2007. Embodied magic lenses and personal territories for wall displays,” [15] B. Lee, P. Isenberg, N. H. Riche, and S. Carpendale, “Beyond mouse in Proceedings of the 2015 International Conference on Interactive and keyboard: Expanding design considerations for information visual- Tabletops & Surfaces. ACM, 2015, pp. 117–126. ization interactions,” IEEE Transactions on Visualization and Computer [36] K.-P. Yee, “Two-handed interaction on a tablet display,” in CHI’04 Graphics, vol. 18, no. 12, pp. 2689–2698, 2012. Extended Abstracts on Human Factors in Computing Systems. ACM, [16] D. Keim, “Information visualization and visual data mining,” IEEE 2004, pp. 1493–1496. Transactions on Visualization and Computer Graphics, vol. 8, no. 1, [37] M. Kaltenbrunner, T. Bovermann, R. Bencina, and E. Costanza, “Tuio: pp. 1–8, 2002. A protocol for table-top tangible user interfaces,” in Proc. of the The [17] R. A. Ruddle, R. G. Thomas, R. S. Randell, P. Quirke, and D. Treanor, 6th Intl Workshop on Gesture in Human-Computer Interaction and “Performance and interaction behaviour during visual search on large, Simulation, 2005, pp. 1–5. high-resolution displays,” Information Visualization, vol. 14, no. 2, pp. [38] M. Fetter and D. Bimamisa, “Tuiofxtoolkit support for the development 137–147, 2015. of javafx applications for interactive tabletops,” in Human-Computer [18] J. C. Roberts, “Exploratory Visualization with Multiple Linked Views,” Interaction. Springer, 2015, pp. 486–489. in Exploring Geovisualization. Amsterdam: Elseviers, 2005, pp. 159– [39] M. Fetter, D. Bimamisa, and T. Gross, “Tuiofx: A javafx toolkit 180. for shared interactive surfaces,” Proceedings of the ACM on Human- [19] M. Czerwinski, G. Smith, T. Regan, B. Meyers, G. G. Robertson, and Computer Interaction, vol. 1, no. 1, p. 10, 2017. G. Starkweather, “Toward characterizing the productivity benefits of very [40] M. Lichman, “UCI machine learning repository,” 2013. [Online]. large displays.” in Interact, vol. 3, 2003, pp. 9–16. Available: http://archive.ics.uci.edu/ml [20] P. Isenberg, P. Dragicevic, W. Willett, A. Bezerianos, and J.-D. Fekete, [41] A. Finkelstein and M. Range, “Image Mosaics,” Princeton University, “Hybrid-image visualization for large viewing environments,” IEEE Computer Science Department, Technical Report TR-574-98, Mar. 1998. transactions on visualization and computer graphics, vol. 19, no. 12, [42] P. Burgh, “Peripheral viewing and simultaneous contrast,” The Quarterly pp. 2346–2355, 2013. Journal of Experimental Psychology 16(3), pp. 257–263, 1964. [21] T. Tsandilas, A. Bezerianos, and T. Jacob, “Sketchsliders: Sketching [43] S. Anstis, B. Rogers, and J. Henry, “Interactions between simultaneous widgets for visual exploration on wall displays,” in Proceedings of the contrast and colored afterimages,” Vision Research, pp. 899–911, 1978. 33rd Annual ACM Conference on Human Factors in Computing Systems. [44] C. Chubb, G. Sperling, and J. Solomon, “Texture interactions determine ACM, 2015, pp. 3255–3264. perceived contrast,” Proc Natl Acad Sci, 86, pp. 9631–9635, 1989. [22] Y. Zhai, G. Zhao, T. Alatalo, J. Heikkilä, T. Ojala, and X. Huang, [45] R. Steinbrecher, “Bildverarbeitung in der praxis,” RST-Verlag. München- “Gesture interaction for wall-sized touchscreen display,” in Proceedings Wien-Oldenburg, 2005. of the 2013 ACM conference on Pervasive and ubiquitous computing [46] B. Shneiderman, “The eyes have it: A task by data type taxonomy for adjunct publication. ACM, 2013, pp. 175–178. information visualizations,” in IEEE Visual Languages, 1996, pp. 336– [23] M. Heilig, S. Huber, M. Demarmels, and H. Reiterer, “Scattertouch: a 343. multi touch rubber sheet scatter plot visualization for co-located data [47] L. Shao, N. Silva, E. Eggeling, and T. Schreck, “Visual exploration exploration,” in ACM International Conference on Interactive Tabletops of large scatter plot matrices by pattern recommendation based on eye and Surfaces. ACM, 2010, pp. 263–264. tracking,” in Proceedings of the 2017 ACM Workshop on Exploratory [24] R. Sadana and J. Stasko, “Expanding selection for information visual- Search and Interactive Data Analytics. ACM, 2017, pp. 9–16. ization systems on tablet devices,” in Proceedings of the 2016 ACM on [48] D. Gordon, J.-H. Hanne, M. Berchtold, A. A. N. Shirehjini, and Interactive Surfaces and Spaces. ACM, 2016, pp. 149–158. M. Beigl, “Towards collaborative group activity recognition using mo- [25] U. Kister, P. Reipschläger, and R. Dachselt, “Multilens: Fluent inter- bile devices,” Mobile Networks and Applications, vol. 18, no. 3, pp. action with multi-functional multi-touch lenses for information visual- 326–340, 2013. ization,” in Proceedings of the 2016 ACM on Interactive Surfaces and Spaces. ACM, 2016, pp. 139–148. [26] K. Matković, H. Abraham, M. Jelović, and H. Hauser, “Quantitative externalization of visual data analysis results using local regression mod- els,” in International Cross-Domain Conference for Machine Learning and Knowledge Extraction. Springer, 2017, pp. 199–218. [27] J. M. Rzeszotarski and A. Kittur, “Kinetica: naturalistic multi-touch data visualization,” in Proceedings of the 32nd annual ACM conference on Human factors in computing systems. ACM, 2014, pp. 897–906. [28] C. Andrews, A. Endert, B. Yost, and C. North, “Information visualization on large, high-resolution displays: Issues, challenges, and opportunities,” Information Visualization, vol. 10, no. 4, pp. 341–355, 2011. 96