information visualization: presentation
TRANSCRIPT
13/03/14 pag. 1
Information visualization lecture 4
presentation
Katrien Verbert Department of Computer Science
Faculty of Science Vrije Universiteit Brussel
13/03/14 pag. 2
Support for report prepara+on. Many sources of content are visible and ready to hand
A problem
13/03/14 pag. 3
The presentation issue
present (tr.v): to offer to view; display.
13/03/14 pag. 4
overview
Space limita+ons
• Scrolling • Overview + detail • Distor+on • Suppression • Zoom and pan
Time limita+ons
• Rapid serial visual presenta+on
• Eye-‐gaze
13/03/14 pag. 5
Space limitations
13/03/14 pag. 6
7.1 A PROBLEM
Many of us have found ourselves with a report that has to be completed by a deadline, with the result (Figure 7.1) that the dining room table, extended to its 12-guest state, is covered by piles of paper as well as reports, books, clippings and slides; perhaps with more arranged on the floor and on a couple of chairs. There may even be piles on top of piles. Such a presentation of vital information makes a lot of sense: everything relevant is to hand (hopefully!) and, moreover, its very visibility acts as a reminder (Bolt, 1984, page 2) of what might be relevant at any particular juncture, possibly triggering a situated action (Suchman, 1987). In this environment I can concentrate on creative tasks rather than organisation.
Despite the availability of high-resolution displays and powerful workstations I still write most of my reports in this way. Why? Because the display area provided by the typical workstation is far too small to support, visibly, all the sources that are relevant to my composition.
7.2 THE PRESENTATION PROBLEM
I am not alone in the sense of having too much data to fit onto a small screen. A very large and expensive screen, for example, would be needed to display the London Underground map in sufficient detail(Figure 1.1), and it would be difficult or impossible to present, on a normal display, the complete organisation chart of IBM or ICI. Moreover, the recent emergence of small and mobile information and communication devices such as PDAs and wearable displays has additionally identified a pressing need for a solution to the ‘ too much data, too little display area’ problem: the presentation problem. How can it be solved, mindful of the need to support the activity of visualising the underlying data?
7.2.1 Scrolling
An obvious solution is to scroll the data into and out of the visible area. In other words, to provide a means whereby a long document can be moved past a window until it reaches the required ‘page’ (Figure 7.2). This mechanism is widely used, but carries with it many penalties. One relates to the "Where am I?" problem: I’m working on Chapter 2, (it may be section 2.3, I don’t know) and I want to remind myself of a figure that is in chapter 5, it may be in section 5.3 – or was it 5.6? All I can do is operate the scrolling mechanism and look out for the figure I need, albeit assisted by various cues such as the page number indicated in the scrolling mechanism. With a scrolling mechanism, most of a document is hidden from view. I have the same problem when using a microfilm reader, with the additional complication that if I move the tray to the left, the image moves to the right. A similar difficulty applies to my use of the famous London ‘AtoZ’ street directory. I’m driving along a road that goes off the edge of the page, so I desperately need whatever page contains the continuation of that road (and quickly!). Even if I get it, I will typically have trouble locating the same road on the new page. These and other similar problems can be ameliorated by the provision of context. Much of this chapter, in fact, is concerned with deciding how to provide context.
Scrolling
13/03/14 pag. 7
Overview + detail
13/03/14 pag. 8
Source: Courtesy Colin Grimshaw
Overview + detail
13/03/14 pag. 9
Overview + detail
hDp://www.datavis.ca/milestones/
13/03/14 pag. 10 A journey north towards Halifax requires detail of the town (Huddersfield) through which the traveller passes
Overview + detail
13/03/14 pag. 11
The use of a real or digitally simulated magnifying glass masks detail around the magnified region
Overview + detail
13/03/14 pag. 12 The DragMag technique allows flexible posi+oning of the region to be magnified
Overview + detail
13/03/14 pag. 13
Connection between the detail and overview presentations missing
Overview + detail
Issues?
13/03/14 pag. 14
Focus + context
13/03/14 pag. 15
Metaphor illustra+ng the principle of the Bifocal Display
(a) An information space containing documents, emails, etc.
(b) The same space wrapped around two uprights.
(c) Appearance of the information space when viewed from an appropriate direction
direction of view
Distortion
13/03/14 pag. 16
An early illustration of the bifocal display principle
13/03/14 pag. 17
An early illustration of the bifocal display principle
13/03/14 pag. 18
Bifocal display features
1. Distortion: available display area is allocated to two different regions – Focus (undistorted) – Context (distorted)
2. Information moves smoothly and continuously from context to focus 3. Display affords for representation
– opportunity to use two dimensions – for instance, +me assigned to horizontal axis – type of item to Y-‐axis
4. Main purpose – Focus: provide detail – Context: awareness and iden6fica6on
5. Manual control
13/03/14 pag. 19
What is the Bifocal Display Doing?
Transforming the information space to the display space
7.19
Informa+on space
Display Space Normal
display
Informa+on space
Display Space
Bifocal display
context
focus
Slide source: Ken Brodlie
13/03/14 pag. 20
A sequence of amino acids within a protein Source: Courtesy of Tom Oldfield
Applications of distortion technique
13/03/14 pag. 21
Table lens without distortion
13/03/14 pag. 22
Table lens with distortion
13/03/14 pag. 23
Table Lens: demo hD
p://www.youtube.com
/watch?v=qW
qTrRAC52U
13/03/14 pag. 24
Schematic representation of X-distortion
13/03/14 pag. 25
Schematic representation of combined X- and Y-distortion
13/03/14 pag. 26
hDp://www.youtube.com/watch?v=D1ediZXIDkc
13/03/14 pag. 27
Distorted presentation of the London Underground map
13/03/14 pag. 28
11Sun
12 Mon
13 Tue
14 Wed
15 Thur
16 Fri
17Sat
Fly LAKathy to airport Model Maker
Check slides, notes.Family barbeque
Fly LHR Kathy to collectChapter 2/ see Dave March
JulyJuneMayAprilMar Aug Sept Oct
Flight to SFOTutorial set-upTutorialUnited flight Heathrow
PointerColor OHsJane+John
Call Kathy
Combined X- and Y-distortion provides a convenient calendar interface
13/03/14 pag. 29
Visual designer’s sketch of the applica+on of the flip-‐zoom technique to the presenta+on of photographs on a Nokia mobile phone Source: Courtesy Ron Bird
13/03/14 pag. 30
Source: Courtesy David Baar, IDELIX SoFware Inc.
Distorted map on a PDA, showing the continuity of transportation links
13/03/14 pag. 31 Source: Courtesy IDELIX and Mitsubishi
Distorted map on a table
13/03/14 pag. 32
Equal X- and Y-distortion centred around a manually chosen location in the Macintosh OSX ‘dock’
13/03/14 pag. 33
The Perspective Wall applies a 3D effect to the bifocal display
13/03/14 pag. 34
Advantages Perspective Wall
• User can adjust ratio of detail to context • Smooth animation helps user perceive object constancy • Relationship between detail and context is consistent: objects
bend around the corner
Slide source: Ken Brodlie
13/03/14 pag. 35
Perspective Wall
Perspective gives smoother transition from focus to context
Informa+on space
Display Space
Perspective Wall
context
focus
Slide source: Ken Brodlie
13/03/14 pag. 36
overview
Space limita+ons
• Scrolling • Overview + detail • Distor+on • Suppression • Zoom and pan
Time limita+ons
• Rapid serial visual presenta+on
• Eye-‐gaze
13/03/14 pag. 37
Suppression
• Applies a distance function and relevance function
• Less relevant other items are dropped from the display
• Classic example: New Yorker’s idea of the world
13/03/14 pag. 38
Suppression
• Originally proposed by Furnas (1986), but many variations of applications.
• Basic idea: more relevant information presented in great detail; the less relevant information presented as an abstraction.
• Relevance is computed on basis of the importance of information elements and their distance to the focus.
13/03/14 pag. 39
Degree of interest (DOI) function:
DOI(a|.=b) = API(a) – D(A,b) • DOI(a|.=b): DOI of a, given the current focus is b. • API(a): sta+c global a priori importance measure. • D(a,b): distance between a and b.
13/03/14 pag. 40
G
P
President
S
M N
F
K
The organization tree of a company
13/03/14 pag. 41
1
2
3 3
4 422
1 11
22
P Focus
Showing the ‘distance’ of each node from the focus of attention
13/03/14 pag. 42
Focus
Context
P
S
M NK
The context defined by setting an upper threshold of unity for distance from a focus
13/03/14 pag. 43
Example of a display that might be associated with the focus and context
13/03/14 pag. 44
Each node in the organiza+on tree has been assigned an a priori importance (API)
10
9 9
8
7 77
8 8
6
8 8
6
9
API
13/03/14 pag. 45
Degree of Interest (DoI)
DoI = API – D Expressed as a function of two quantities: • A priori importance (API) • Distance (D) between an item and the item currently in focus
13/03/14 pag. 46
Segng a lower limit of 6 for DoI iden+fies the nodes within the shaded region
8
6 6
8
6 66
4 4
4
6 6
4
8
Focus
Context
Nodal values of degree of interest (=API – D)
13/03/14 pag. 47
Part of an engineering drawing
Applications of DoI concept
13/03/14 pag. 48
The engineering drawing simplified in the context of a suspected fault
Applications of DoI concept
13/03/14 pag. 49
Illustra+ng the concept of a magic lens. (a) shows a conven+onal map of an area, (b) shows the loca+on of services (gas, water and electricity pipes) in the same area, and (c) a (movable) magic lens shows services in an area of interest, in context
Application in magic lens technique
13/03/14 pag. 50
hDp://www.youtube.com/watch?v=2bYDKbzocSg
13/03/14 pag. 51
A molecular surface of the protein transferase coloured by electrosta+c poten+al bound to DNA shown as a schema+c. (ID = 10mh). The magic lens window allows a view of the atomic structure bonding to be shown, with the bound ligand structure highlighted as cylinders, thereby providing a view inside the protein Source: By kind permission of Tom Oldfield and Michael Hartshorn
Magic lens
13/03/14 pag. 52
A 3D Flexible and Tangible Magic Lens in Augmented Reality
www.youtube.com/watch?v=PKegByAZ0kM
13/03/14 pag. 53
A combina+on of rubber-‐sheet distor+on and suppression lead to a map appropriate to a journey from one city to another
Combined distortion and suppression
13/03/14 pag. 54
The rubber-sheet distortion technique employed in the map
13/03/14 pag. 55
Historical note
• Distortion and suppression appeared in early 1980s • Need to maintain a balanced view of focus + context
identified earlier – for example by Farrand (1973)
“an effective transformation must somehow maintain global awareness while providing detail”
“… there is a need for presenting a display with 1. sufficient
detail for interaction, while 2. maintaining global vision of the entire scene.”
13/03/14 pag. 56
Fisheye view
• Farrand also coined the term “fisheye” • Nowadays appears to refer to both distortion and suppression
13/03/14 pag. 57
Fisheye Menus
• Here is the same idea applied to menus – Ben Bederson, University of Maryland
• See also: – hDp://www.cs.umd.edu/hcil/fisheyemenu/fisheyemenu-‐demo.shtml
ENV 2006
13/03/14 pag. 58
Fisheye View, Polyfocal Display
Can distort boundaries because applied radially rather than x y
1D Fisheye 2D Polyfocal
Slide source: Hornung and Zagreus
13/03/14 pag. 59
hDp://w
ww.cs.um
d.edu/class/fall2002/cm
sc838s/+chi/fisheye.html
13/03/14 pag. 60
hDp://www.youtube.com/watch?v=dEowKzbDpKU
hDps://plus.google.com/+MiguelRodriguez/posts/YqBm18xQgQc
13/03/14 pag. 62 Source: By kind permission of Patrick Baudisch
The use of representation (by a ‘halo’) to provide context for a small display
13/03/14 pag. 63
overview
Space limita+ons
• Scrolling • Overview + detail • Distor+on • Suppression • Zoom and pan
Time limita+ons
• Rapid serial visual presenta+on
• Eye-‐gaze
13/03/14 pag. 64 Panning is the smooth movement of a viewing frame over a 2D image
Zoom and pan
13/03/14 pag. 65 Zooming is the increasing magnifica+on of a decreasing frac+on of an image (or vice versa)
Zoom and pan
13/03/14 pag. 66
Zooming
• Conventional zooming-in – No change in data or representa+on – only filtering – Loss of context
• ≠distortion whose purpose is to permit focusing rather than filtering
• Supports two cognitive tasks (Cairns and Craft 2005) – Zooming-‐in: extraneous informa+on is removed from visual field – more
manageable view – Zooming-‐out: reveals hidden informa+on
13/03/14 pag. 67
A space-scale diagram relevant to combined zooming and panning
Furnas and Bederson (1995)
13/03/14 pag. 68
Google earth
13/03/14 pag. 69
Exploring Information Spaces by Using Tangible Magic Lenses
hDp://w
ww.youtube.com
/watch?v=h-‐m
F4_OAhU
0
13/03/14 pag. 70
Semantic zoom
• Previous example: geometric zoom – Con+nuous – Zooming-‐in: filtering and loss of context
• Semantic zoom – Discrete transi+on – Addi+onal detail
13/03/14 pag. 71
A combination of geometric and semantic zoom
13/03/14 pag. 72
overview
Space limita+ons
• Scrolling • Overview + detail • Distor+on • Suppression • Zoom and pan
Time limita+ons
• Rapid serial visual presenta+on
• Eye-‐gaze
13/03/14 pag. 73
A collec+on of images is presented, one at a +me, at a rapid rate (e.g., ten per second)
time
Rapid serial visual presentation
13/03/14 pag. 74
Tile mode: concurrent presentation of images opposed to ‘slide show mode’
13/03/14 pag. 75
‘Floa+ng RSVP’ in which images appear to approach the viewer from a distance. Sensi+ve arrows allow the speed and direc+on of ‘movement’ to be controlled by a user Source: Courtesy Kent WiNenburg
Floating RSVP
13/03/14 pag. 76
The contents of an online bookstore are presented in ‘collage mode’ RSVP, simula+ng the placing of book covers on a table in sequence. The set of arrows just under the presenta+on allows control of the speed and direc+on of presenta+on Source: Courtesy Kent WiNenburg
Collage mode RSVP
13/03/14 pag. 77
An interface facilita+ng the browsing of posters adver+sing videos. Cursor movement along the stacks causes posters to briefly ‘pop out’ sideways, and the whole bifocal structure can be scrolled to bring a video of interest to the central region, where a mouse click will cause a clip from a video to be played (Lam and Pence 1997)
RSVP + bifocal principle
13/03/14 pag. 78
Space-time trade-off
13/03/14 pag. 79
Space-time trade-off
13/03/14 pag. 80
An experiment to test a subject’s ability to recognise the presence or absence of a previously viewed target image within a collec+on presented sequen+ally at a rate of around ten per sec.
Prior instruction to subject
Subjectsʼ performance
“Here is a target image. Tell me if this image appears in the sequence of N images youʼre about to see”
Recognition about 80% to 90% successful
time
about 100 ms
unrelated images
Presentation of images
Briefly glimpsed images
13/03/14 pag. 81
Representa+on of limits on display area and total presenta+on +me by a ‘resource box’
Display area
Presentationtime
Space and time resources
13/03/14 pag. 82 Source: Courtesy of Katy Cooper
Three ‘static’ image presentation modes (A, B, C) and three ‘moving’ image presentation modes (D, E, F)
13/03/14 pag. 83
Slide show
13/03/14 pag. 84
Mixed
13/03/14 pag. 85
Tile
13/03/14 pag. 86
Diagonal
13/03/14 pag. 87
Ring
13/03/14 pag. 88
Stream
13/03/14 pag. 89 Source: Courtesy of Katy Cooper
Favorite mode?
13/03/14 pag. 90
The accuracy with which the presence or absence of a target image was reported for the six presenta+on modes, averaged over all tasks and presenta+on +mes.
1.00.90.80.7
0.60.50.40.3
0.20.1
Slide-show Mixed Tile Diagonal Ring Stream
Recognition accuracy
Presentation modes
13/03/14 pag. 91 The (sta+c) slide-‐show, mixed and +le image presenta+on modes account for three-‐quarters of the preferred modes (Cooper et al. 2006)
13/03/14 pag. 92
Almost all the least preferred image presenta+on modes were moving modes and the stream mode accounted for over half
13/03/14 pag. 93
A simple representa+on of eye-‐gaze behaviour. The rapid saccades are shown green, the fixa+ons (F) of varying dura+on by circles of propor+onate size
F
F
F
F
FF F
F
Eye-gaze
13/03/14 pag. 94
Eye-gaze trajectory slide show
13/03/14 pag. 95
Eye-gaze trajectory tile mode
13/03/14 pag. 96
Eye-gaze trajectories mixed mode
13/03/14 pag. 97
Eye-gaze trajectories diagonal mode for subject who disliked the mode
13/03/14 pag. 98
Eye-gaze trajectories diagonal mode for subject who liked the mode
13/03/14 pag. 99
Eye-gaze trajectory stream for subject who disliked the mode
13/03/14 pag. 100
The acquisi+on of an expanding target. (a) The dormant appearance of the image collec+on, and (b) its appearance when the cursor rests over image 6
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
10 11 12 13 14 15 16 17 18 19 209861 2 3 4
(a)
(b)
Manual control: ‘expanding target’ presentation mode
13/03/14 pag. 101
An experiment in which a subject first views the rapid (e.g., 10 per second) presenta+on of a collec+on of images, is then shown a single image and asked to say whether that image was part of the collec+on. Iden+fica+on success is highly dependent upon the +me elapsing between the end of the presenta+on and the ques+oning of the subject
Prior instruction
to subjectPresentation of image collection Subject’s performance
about 100ms
unrelated images
time
None
The subject was shown an image and then
asked, ‘Was this image present in the
sequence you have just seen?’
Recognition success was 10% to 20%
unless the question was aksed within about
4 seconds of the end of the presentation
Models of human visual performance
13/03/14 pag. 102
An experiment in which a collec+on of images is presented to a subject. Each image is presented briefly (e.g., for 100ms) and followed by a ‘visual mask’ las+ng about 300ms. Subjects were able to say, with a considerable degree of success, whether an image shown arerwards had been part of the presenta+on
Prior instruc+on to subject
Presenta+on of image collec+on Subject’s performance
about 300ms
unrelated images +me
None
The subject was shown an image and then asked, ‘Was this image present in the sequence you have just seen?’
Up to 92% recogni+on success etc . . . .
Visual mask .
Visual mask
Visual mask
about 100ms
Models of human visual performance
13/03/14 pag. 103
A third palleDe for the interac+on designer, addressing issues of presenta+on
Presentation concepts and techniques Scrolling
Overview+detail
Distortion
Suppression
Zoom Pan
RSVP Eye gaze
Recap
13/03/14 pag. 104
Visual Information Seeking: Mantra
Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand Overview, zoom & filter, details-on-demand
Ben Shneiderman, 1996
13/03/14 pag. 105
Shneiderman’s “7 Tasks”
• Overview task – overview of entire collection
• Zoom task – zoom in on items of interest
• Filter task – filter out uninteresting items
• Details-on-demand task – select an item or group to get details
• Relate task – relate items or groups within the
collec+on
• History task – keep a history of ac+ons to support
undo, replay, and progressive refinement
• Extract task – allow extrac+on of sub-‐collec+ons and
of the query parameters
13/03/14 pag. 106
Questions?
13/03/14 pag. 107
Readings
• Chapter 4
13/03/14 pag. 108
References
• Furnas, G. W., & Bederson, B. B. (1995, May). Space-scale diagrams: Understanding multiscale interfaces. In Proceedings of the SIGCHI conference on Human factors in computing systems (pp. 234-241). ACM Press/Addison-Wesley Publishing Co..
• Shneiderman, B. (1996, September). The eyes have it: A task by data type taxonomy for information visualizations. In Visual Languages, 1996. Proceedings., IEEE Symposium on (pp. 336-343). IEEE.
• Some relevant notes:http://jcsites.juniata.edu/faculty/rhodes/ida/presentation.html
13/03/14 pag. 109
Research presentations
13/03/14 pag. 110
Research presentations
• Schedule on PointCarré
• Select a second paper in the same slot for questions: e.g. session 1: http://doodle.com/rmpc9g8u3p2qzsy4
• Links to doodle polls for all six sessions will be included in the schedule.
13/03/14 pag. 111
Team project
13/03/14 pag. 112
Team project milestones
1. Form teams 2. Project proposal 3. Intermediate presentation 4. Final presentation 5. Short report
due 27 Feb.
due 13 March
due 3 April
22 May
due 29 May
13/03/14 pag. 113
Project proposal
1 page description of your intended project: – mo+va+on – which datasets you will use – current status. If available, first designs. – problems/ques+ons
due 13 March If you want earlier feedback, send us your proposal earlier ;-)
13/03/14 pag. 114
Data collection
• https://docs.google.com/forms/d/1gHwVWHZLzWdSz1F37jA1Gungrl56bT215M6FYW3YqGY/viewform Or
• bit.ly/N6JTyD
Anonymous! Choose your own ID.
• Please report your data ;-)