1997 International Co,nference on
`Intelligent User Interfaces
`
`Editors
`Johanna Moore, Papers & Program Chair
`Ernest Edmonds, Panels
`Angel Puerta, Papers & Debate
`
`Sponsored by:
`ACM SIGART - Special Interest Group on Artificial Intelligence
`ACM SIGCHI - Special Interest Group on Computer-Human Interaction
`
`IUl97 January 6-9, 1997 - Orlando, Florida, USA
`
`1 of 15
`
`

`

`The Association for Computing Machinery
`1515 Broadway
`New York, N.Y. 10036
`
`Copyright <0 1997 by the Association for Computing Machinery, Inc.(ACM). Permission to
`make digital or hard copies of all of this work for personal or classroom use is granted without
`fee provided that the copies are not made or distributed for profit or commercial advantage
`and that copies bear this notice and the full citation on the first page. Copyrights for compo(cid:173)
`nents of this work owned by others than ACM must be honored. Abstracting with credit is
`permitted. To copy otherwise, to republish, to post on servers or to redistribute to lists
`requires prior specific permission and/or a fee. Request permission to republish from: Publi(cid:173)
`,.,;
`cations Dept. ACM, Inc. Fax + 1 (212) 869-0481 or <permissions@acm.org>
`For other copying of articles that carry a code at the bottom of the first or last page, copying
`is permitted provided that the per-copy fee indicated in the code is paid through the Copyright
`Clearance Center, 222 Rosewood Drive, Danvers, MA 01923.
`
`ACM ISBN: 0-89791-839-8
`
`Additional copies may be ordered prepaid from:
`
`ACM Order Department
`PO Box 12114
`Church Street Station
`New York NY 10257
`
`ACM European Service Center
`108 Cowley Road
`Oxford OX 4 lJF UK
`Phone: +44-1-865-382338
`Fax: +44-1-865-3811338
`E-mail: acm_europe@acm.org
`URL: http://www.acm.org
`
`ACM Order Number: 608970
`
`Phone: 1-800-342-6626
`(USA and Canada)
`+ 1-212-944-1318
`Fax: + 1-21-944-1318
`E-mail: acmpubs@acm.org
`
`Printed in the U.S.A.
`
`2 of 15
`
`

`

`TABLE OF CONTENTS
`
`PLENARY ADDRESS
`
`Technology Transfer from University to Industry
`James Foley, Mitsubishi Electric Research Laboratory (U.S.A.) ............................. 3
`
`PAPERS I: Planning Based Approaches
`
`Local Plan Recognition in Direct Manipulation Interfaces
`Annika Wrern (Sweden) ........................................................................... 7
`
`Interaction with a Mixed-Initiative System for Exploratory Data Analysis
`Robert St. Amant and Paul R. Cohen (U.S.A.) ................................................ 15
`
`Segmented Interaction History in a Collaborative Interface Agent
`Charles Rich and Candace L. Sidner (U.S.A.) ................................................. 23
`
`DEBATE: Direct Manipulation vs. Interface Agents
`
`Direct Manipulation for Comprehensible, Predictable and Controllable
`User Interface
`Ben Shneiderman (U.S.A.) ....................................................................... 33
`
`Intelligent Software
`Pattie Maes (U.S.A.) .............................................................................. 41
`
`3 of 15
`
`

`

`PAPE RS II:
`
`Interface Agents
`
`The Selection Recognition Agent: Instant Access to Relevant
`Information and Operations
`Milind S. Pandit and Sameer Kalbag (U.S.A.) ................................................. 47
`
`Using Agents to Personalize the Web
`Christoph G. Thomas (Germany) and Gerhard Fischer (U.S.A.) ............................ 53
`
`Multimodal User Interfaces in the Open Agent Architecture
`Douglas B. Moran, Adam J. Cheyer, Luc E. Julia, David L. Martin (U.S.A.)
`Sangkyu Park (Korea) ............................................................................. 61
`
`PAPERS III: Presentation Aids/Coordination
`
`Generating Web-Based Presentations in Spatial Hypertext
`Frank M. Shipman, ill, Richard Furuta, Catherine C. Marshall (U.S.A.) .................. 71
`
`Adding Animated Presentation Agents to the Interface
`Thomas Rist, Elisabeth Andre, Jochen Muller (Germany) .................................... 79
`
`Dynamic Dramatization of Multimedia Story Presentations
`Nikitas M. Sgouros, George Papakonstantinou,
`Panagiotis Tsanakas (Greece) ..................................................................... 87
`
`PAPE RS IV: 1/0 Support/Spatial Awareness
`
`Description and Recognition Methods for Sign Language Based on
`Gesture Components
`Hirohiko Sagawa, Masaru Takeuchi, Masaru Ohki (Japan) ................................... 97
`
`Haptic Output in Multimodal User Interfaces
`Stefan Munch and Rudiger Dillmann (Germany) ............................................. 105
`
`Helping Users Think in Three Dimensions: Steps Toward Incorporating
`Spatial Cognition in User Modelling
`Michael Eisenberg, Ann Nishioka, M. E. Schreiner (U.S.A.) ............................ 113
`
`II
`
`~~~--------------------................ .
`
`4 of 15
`
`

`

`PANEL I: Computational Approaches to Interface Design: What
`Works, What Doesn't, What Should and What Might
`
`Organizer: Christopher A. Miller (U.S.A.)
`Panelists: Kevin Corker (U.S.A.)
`Mark Maybury (U.S.A.)
`Christopher A. Miller (U.S.A.)
`Angel R. Puerta (U.S.A.) ...................................................... 123
`
`PAPERS V: Automation of Presentations
`
`Top-Down Hierarchical Planning of Coherent Visual Discourse
`Michelle X. Zhou and Steven K. Feiner (U.S.A.) ........................................... 129
`
`Declarative Models of Presentation
`Pablo Castells (Spain), Pedro Szekely (U.S.A.), Ewald Salcher (Austria) ............... 137
`
`Integrating Planning and Task-based Design for Multimedia Presentation
`Stephan Kerpedjiev, Giuseppe Carenini, Steven F. Roth,
`Johanna D.Moore (U.S.A.) ..................................................................... 145
`
`PAPERS VI: Applications
`
`The Pedagogical Design Studio: Exploiting Artifact-Based Task Models
`for Constructivist Learning
`James C. Lester, Patrick J. FitzGerald, Brian A. Stone (U.S.A) ......................... 155
`
`Some Interface Issues in Developing Intelligent Communications Aids
`for People with Disabilities
`Kathleen F. McCoy, Patrick Demasco, Christopher A. Pennington,
`Arlene LuberoffBadman (U.S.A.) ............................................................ 163
`
`iii
`
`5 of 15
`
`

`

`PANEL II: Compelling Intelligent User Interfaces:
`How Much AI?
`
`Organizer:
`Panelists:
`
`Joe Marks (U.S.A.)
`Larry Birnbaum (U.S.A.)
`Eric Horvitz (U.S.A.)
`David Kurlander (U.S.A.)
`Henry Lieberman (U.S.A.)
`Steven Roth (U.S.A.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
`
`PAPERS VII: Web /Hypermedia
`
`Evaluating the Utility and Usability of an Adaptive Hypermedia System
`Kristina Hook (Sweden) ......................................................................... 179
`
`Multi-Level User Support through Adaptive Hypermedia: A Highly
`Application-Independent Help Component
`L. Miguel Encarna~ao (Germany) ..................................... ......................... 187
`
`Decision Making in Intelligent User Interfaces
`Constantine Stephanidis, Charalampos Karagiannidis,
`Adamantios Koumpis (Greece) ................................................................. 195
`
`PLENARY ADDRESS
`
`What Makes an Intelligent User Interface Intelligent?
`Doug Riecken, Bell Laboratories (U.S.A.) ................................................... 205
`
`SHORT PAPERS
`
`An Adaptive Short List for Documents on the World Wide Web
`Matjaz Debevc (Slovenia), Beth Meyer (U.S.A.), Rajko Svecko (Slovenia) ............. 209
`
`An Interface Agent for Nonroutine Tasks
`Yuzo Fujishima ................................................................................... 213
`
`An Interface for Collaborative and Coached Approaches to Learning
`Critical Inquiry
`Dan Suthers and the Adv learn Project (U.S.A.) .. .. ... ........................................ 217
`
`IV
`
`6 of 15
`
`

`

`Automating a Classification Task Based on an Augmented Thesaurus
`Eunok Paek and Hye-Jeong Jeon (Korea) ..................................................... 221
`
`Easing Interaction through User-Awareness
`Alain Karsenty (France) ......................................................................... 225
`
`Individual User Interfaces and Model-based User Interface
`Software Tools
`Egbert Schlungbaum (Germany) ............................................................... 229
`
`Inductive Task Modeling for User Interface Customization
`David Maulsby (U.S.A.) ....................................................................... 233
`
`Intelligent Network News Reader
`Hitoshi !sahara and Hirorni Ozaku (Japan) ................................................... 237
`
`Intelligent Word-Prediction to Enhance Text Input Rate
`(A Syntactic Analysis-Based Word-Prediction Aid for People with Severe
`Motor and Speech Disability)
`Nestor Garay-Vitoria and Julio Gonzalez-Abascal (Spain) .................................. 241
`
`Interactive Model-based Coding for Face Metaphor User Interface
`in Network Communications
`Kazuo Ohzeki, Takahiro Saito, Masahide Kaneko and
`Hiroshi Harashima (Japan) ...................................................................... 245
`
`Management of Interface Design Knowledge with MOBI-D
`Angel R. Puerta and David Maulsby (U.S.A.) ................................................ 249
`
`Providing User Support for Interactive Applications with FUSE
`Frank Lonczewski (Germany) .................................................................. 253
`
`A Response Model for a CG Character Based on Timing of
`Interactions in a Multimodal Human Interface
`Kenji Sakamoto, Haruo Hinode, Keiko Watanuki, Susumu Seki,
`Jiro Kiyama, Furnio Togawa (Japan).......................................................... 257
`
`The Stick-e Note Architecture: Extending the Interface Beyond the User
`Jason Pascoe (U.K.) .............................................................................. 261
`
`Wizards, Guides, and Beyond: Rational and Empirical Methods for
`Selecting Optimal Intelligent User Interface Agents
`D. Christopher Dryer (U.S.A.) ................................................................. 265
`
`INDEX
`
`Author Index ....................................................................................... 271
`
`v
`
`7 of 15
`
`

`

`GENERATING WEB-BASED PRESENTATIONS
`IN SPATIAL HYPERTEXT
`
`Frank M. Shipman III, Richard Furuta, Catherine C. Marshall*
`
`Department of Computer Science & Center for the Study of Digital Libraries
`Texas A&M University
`College Station, TX 77843-3112
`+l 409 862 3216
`E-mail: {shipman, furuta, marshal I }@cs.tamu.edu
`
`ABSTRACT
`Presentations frequently include material appropriated from
`external sources; they may incorporate tabular data from
`published reports, photographs from books, or clip art from
`purchased collections. With the growing use of the World(cid:173)
`Wide Web to dis . eminate information, there is the emerging
`potential for a new style of presentation: one that interprets
`and organizes materials produced by others and published
`on-line. Authoring such presentation requires the analysis
`of the source information. However, current presentation
`authoring
`' Oftware
`is designed
`to support
`traditional
`presentations, where analysis is assumed a separate task at
`best supported by separate software. This paper di sc usses
`experiences with using VIKI, a system designed to support
`information
`analy is,
`for
`the
`authoring of
`such
`presentations. VIKI includes a spatial parser to recognize
`implicit spatial structure generated during analysis. This
`paper describe · how initial experiences with use for path
`authoring
`led
`to VIKI enhancements,
`including
`the
`adaptation of implicit spatial structure recognition for the
`creation of presentations.
`
`KEYWORDS: Presentation Authoring, Analysis TooL ,
`Spatial Parsing, Implicit Structure, Presentation Models,
`Spatial Hypertext, World-Wide Web, Walden's Pathq VIKI
`
`INTRODUCTION
`Creating presentations is a common activity. Teachers,
`cholars, and many business people pend a significant part
`of their
`time producing materials
`they will use
`to
`disseminate information, ideas, and views. Different types
`of presentations rely on background materials taken from
`different source . Teachers traditionally gather materials
`from textbooks and curriculum guides. Scholar reflect on
`published results as well as on the outcomes of their own
`research. Busines people may use market analyses, trade
`
`Pem1ission to make digital/hard copies of all or part of this material for
`personal or classroom use is granted without fee provided that the copies
`are not made or distributed for profit or commercial advantage, the copy(cid:173)
`right notice, the title of the publication and its date appear, and notice is
`given that copyright is by pennission of the ACM, Inc. To copy otherwise,
`to republish, to post on servers or to redistribute to lists, requires specific
`pennission and/or fee.
`IUI 97, Orlando Florida USA
`e 1997 ACM 0-89791-839-8/96/01
`
`.. $3 .50
`
`publications, and internal reports as the basis for their
`presentations.
`
`The common element of all these characterizations is that
`much of the information underlying the presentations is
`derived from paper-based sources. But this is changing.
`More and more new papers, magazines, and academic and
`trade publications are becoming available on-line. These on(cid:173)
`line resources frequently include new facilities that make
`them
`better
`suited
`for
`creating
`information-rich
`presentations: for example, the external
`information is
`readily accessible through full-text search and alternative
`indexing schemes, and document components are easy to
`reuse , given their manipulable electronic representation
`
`like Yahoo and OpenText are
`Information services
`examples of the growing number of businesses providing
`better on-line
`access
`to
`information. Professionally
`generated and edited information is becoming available as
`the World-Wide Web is commercialized. Many newspapers ,
`magazines, and edited academic journals now have on-line
`subscription ervices. Of course, the Web is also a source of
`unedited opinions, conjecture, and gossip. How does this
`ready availability of external sources affect the needs of and
`possibilities for presentation authors?
`
`Most presentation software, such as Microsoft PowerPoint,
`Aldus Persuasion, or Adobe Premiere, exists as stand-alone
`application ~ . This type of software assumes that an author
`creates a self-contained presentation in which the content i
`produced strictly in service of the presentation itself. But the
`growth
`in on-line
`information
`resources suggests
`the
`development of new presentation genres, e .g., forms that
`draw on materials from external ources and reuse content
`in new, unanticipated context..
`
`One alternative model for presentations that we have been
`investigating generalizes Zellweger's "directed paths" (7]
`
`* Author's current address is: Xerox PARC, 3333 Coyote
`Hill Road , Palo Alto, CA 94304. Phone: (415) 812 - 4288,
`E-mail: mar hall@parc.xerox.com
`
`71
`
`8 of 15
`
`

`

`and Trigg's "guided tours" [6]. As originally defined, a
`directed path provided the means for guiding a reader's
`traversal along a path of components extracted from a set of
`documents or an existing hypertext network. The ordering
`of components on the path
`is not constrained by the
`structure of the source documents -- in other words, the
`components do not have to follow the temporal orderings of
`the source. The directed path defines a meta-structure that is
`layered on
`top of the underlying documents' existing
`structures.
`
`This work is also reminiscent of Vannevar Bush's vision of
`a scholar blazing trails through reference materials in a
`memex [I]. The trails that Bush describes are coherent
`linear paths through collected materials, with possible
`digressive
`side
`trips. Rather
`than emphasizing
`the
`expressivene s of the path mechanism as is the focus of
`Zellweger's and Trigg's subsequent work, Bush describes a
`trail as analogous to gathering together physical items from
`widely separated sources; having simple mechanisms for
`flipping through the materials and commenting on them;
`and being able to transmit this annotated linear trail to a
`colleague.
`
`Walden's Paths, our variation of a directed path facility,
`supports the presentation of an annotated linear path of
`Web-based information [5]. In
`these presentations, the
`author creates a
`list of Web pages and associates
`them. By adding some text or other
`annotations with
`annotations to the content of a Web page, the path author
`may provide a rhetorical structure to the path as a whole,
`in gaps between pages, and
`create
`transitions
`to fill
`emphasize particular aspects of the materials.
`
`to
`teachers
`to allow
`is
`The goal of Walden 's Paths
`incorporate Web-based information into their curriculum.
`for students
`is
`Creating a Web-derived presentation
`different than creating a presentation for a conference or
`other professional presentation. Most notably, educational
`paths typically take information authored by others and
`present it within a new context. Authoring such a path is
`then a combination of locating relevant Web sites, selecting
`which materials to include, and creating the context in
`which the materials will be presented to the students.
`
`Constructing coherent paths of this sort involves many
`activities outside
`the scope of standard presentation
`software. Path authors must engage
`i.n
`information
`discovery, validation, and management. For this reason we
`have been investigating the use of VIKI, a system designed
`to support information analysis (3), for the new task of
`authoring paths.
`
`In this paper we first will discuss some of our early
`experiences with the authoring of Web path presentations
`within VIKl. Then we will describe VIKl in more detail and
`the extensions made to VIKI to better ·upport this type of
`authoring process. We focus briefly on how VIKl's
`recognition of implicit spatial structures can aid in the
`authoring of presentations. We conclude with a discussion
`
`of the initial experiences with the modified system, some
`lessons learned, and topics for future work.
`
`INITIAL EXPERIENCES WITH PATH AUTHORING
`In the course of our work on the Walden's Paths mechanism
`we have gained experience with authoring Web-based
`presentations. Thi · experience shows that path authoring is
`a complex task which includes:
`•
`locating promising Web sites,
`• browsing and evaluating materials at these sites,
`• selecting information elements for use in the path,
`• developing an outline for the presentation,
`• placing pages within the sections of the path, and
`• writing the introductory text and annotating the pages.
`Not all of these activities occur during every authoring task;
`nor is the list above intended to imply a strict sequence of
`actions. Rather, an authoring process may require many
`iterations through these commonly identifiable subtasks.
`
`In determining if a page is to be included in the path and the
`role it will fill, the path author must decide whether the
`informational content is appropriate for the presentation.
`One common consideration for appropriateness is accuracy.
`When it is difficult to assess the content accuracy, the
`its validity.
`information source may help determine
`Information from government laboratories, on-line editions
`of journals, and newspapers are apt to be treated differently
`from information from a political party or special-interest
`group. Consideration ~ such as these require analysis as well
`as synthesis.
`
`Current tools for the Web, especially browsers and earch
`engines, focus on the act of locating information and getting
`back to it once it ha been found (for example, through
`bookmarks). Current bookmark organization tools support a
`limited expression of the interpretations resulting from a
`user's analysis -- usually through the use of folders or
`similar mechanisms supporting classification hierarchies,
`but these tools do not upport the process of analysis. In
`addition bookmark lists generally are not intended to aid in
`the public task of information presentation.
`
`We have been building VIKI, a spatial hypertext system to
`aid the analy is of information [3]. Spatial hypertext relies
`on proximity and other patterns of vi · ual attributes to show
`relationships and linkages; most hypertext mechanisms, like
`the Web, rely on point-and-click
`those supported on
`traversal. For example, in the simplest case, two adjacent
`nodes in a spatial hypertext may be said to be linked. Other
`affordances of space are used to define different kinds of
`hypertextual
`·tructure (like hierarchies and composite
`nodes) . VIKl provides a hierarchy of two-dimensional
`spaces, called collections, for organizing symbols
`that
`information objects. Symbols may point
`to
`represent
`information objects that are stored in VIKI's own database,
`or symbols may point to external information like Web
`pages. This ability
`to organize and
`interpret external
`materials is crucial to the application described in this paper.
`
`72
`
`9 of 15
`
`

`

`Ylkt en"-p2-vlk1
`l;'<"j
`Flle Edit Text Graphics Zoom Colors Types Composites WWW
`
`1 1 ~~ fDQJQJ-2.j QJf!.lf!l.QlgJQJ ajf!J Fd.=l=:J
`
`~
`
`Web Sources
`
`Tille: Mop ofEndangorodSpeca h US
`
`Tille: U Midt. iltdupnd Speca re
`
`Path for students
`l
`i ~HID~ Spoodes
`I Tille: n. ..... ,.. milliolll of dilYa-.n tplaataad M.imaJ rpeca Oil hrfl today. e. t 11t.... ,
`
`were rnoreyestmb.y,and tomCITOI' tterevillbeew11. fet.fer.
`
`i:
`
`Tille: Tri>aetB~ pap
`
`Tille: Galaxy FAdanpnd Speca !'.p
`
`-----
`
`Tille: TOFS litforW1alD.
`Tille: Middlib•ry eon.p sm.
`
`Tille: Sou.th-W"t Tenr Local iltdanpnd Sp.a..
`
`Tille: T"""' iltdan pnd Speca
`
`Tille: r.t.ro.mmta! Teduui:als.m... Company (Wellondr)
`
`Tille: TOXIU Resou"" Guido to FAdan pnd Speca
`
`Tille: TOFS I map Gall!ry
`
`Tille:~-lileEda~IEnuRes~
`
`F
`
`F.a>NetSile
`
`t Tiiie: COllrtn'OtMvlow ofFAdupndSpedoo A.ct
`
`I Tille: Sea Turtle ;. T.,..,
`l Tille: Bnwn Peba
`
`Wludir kappenng? and vhy?
`
`It itilnportant:> and can vedo111ythlttglboatiC
`
`TEXT: hup~""vnilhka~""danpndJ>tr.I
`
`"~ Specimln 1 ' -
`
`Tlla: Abroda maaocarpa !map
`-
`Tille: s .. Tw.rtlo;. T"""'
`
`T!llo: Brovn ""1bn
`
`I
`
`I
`
`TEXT: A.1ofOctober1994, ,.,.,...,.,.. 71 threatat<d .,.d "'dangorod rp.a..
`lilted by tho US Firlt .,.d Wildlife ren>lco tr T..,...
`
`\
`I
`
`I
`
`...
`
`I e-u,,,llftltmul PG Uc.>
`
`,,._, r...iNet Silo
`
`I
`
`Title: C....'""M,,;,,., of LulM Jllftd. Sp.0.. A.ct
`URL: hup:i,..,.,.. ... crdm!l.comN m1Valoolhcla.html1Spedoo
`
`~
`
`P-blt rlltm9
`TilJo: Lutpamapr plpot
`URL: hup:iiWayJidtt.tamu~etq,priMarttta.p
`
`l
`
`I
`
`·1
`
`'
`I
`J:
`
`.....
`
`:J
`'
`
`~
`
`-
`
`·-
`
`~-- - - - -
`
`Figure I: Use of VIKI to collect information for a Web-based presentation on endangered species in Texas.
`
`Experience_· with VIKI have shown that users find the
`arrangement of information in space to be a natural method
`of describing the evolving interpretations that are part of
`analysis.
`
`Beca use of the need to collect and analyze Web-based
`information during path authoring, VIKI was used during
`the authoring of severa l paths. VIKI' s
`facilities for
`capturing, organi zing, and interpreting information are well
`matched to the initial
`tages of path authoring outlined
`above. In these initial uses of VIKI for path authoring, users
`built lists, outlines and similar structures within VIKI spaces
`to represent the final path of information.
`
`Figure I shows one such layout generated during the
`creation of a path about endangered species found in Texas.
`The left side of the figure, labeled "Web Sources", shows
`references to We b pages collected by the author as being
`potenti ally of use within the path . The right
`ide of the
`figure, "Path for students", contains labels representing an
`
`outline for the presentation and the Web page references
`placed within the outlined subsections.
`
`At the time this information was collected and annotated,
`there was no mechanism for turning the organization of
`symbols in a VIKI space into a final presentation. After
`using VIKI to locate, interpret, select, and order information
`for a path, the author had to move into another i.nterface to
`generate the path that would be used for presentation.
`
`Obviously, a more seamless connection between VIKI and
`the path presentation mechani m was desirable. In an ideal
`situation, VIKI would produce the path representation used
`by the presentation mechanism, but this was not directly
`possible. The VIKI users were not specifying an explicit
`path within YIKI, but creating a spatial layout that was used
`to
`indicate implicitly , at least tQ the author, a
`linear
`sequence of elements for the path.
`
`Layouts such as that in Figure l suggested the need to take a
`layout of information within VIKI and have the system
`
`73
`
`10 of 15
`
`

`

`items.
`linear ordering of the constituent
`determine a
`Furthermore, in this early case, not all the objects within a
`structure were meant to be pages on the path; some were
`intended a
`labels identifying subsections or notes the
`author made during the authoring process, for example the
`darker colored labels on the right side of Figure I.
`
`In the next section we discuss VIKI in more detail and
`describe modifications made to more completely support
`presentation authoring, including locating the appropriate
`materials, and turning an implicitly structured path into a
`realized presentation.
`
`INTEGRATING PRESENTATION AUTHORING
`INTO VIKI
`VIKI is a system designed to support information analysis.
`Prior experience with Aquanet [2] indicated that users found
`it difficult to explicitly represent their interpretations during
`representation
`Instead of using Aquanet's
`analysis.
`mechanisms
`to specify relations , users placed related
`information objects next to one another in familiar spatial
`patterns like lists and .·tach. VIKI was designed to facilitate
`and enhance this use of spatial arrangement during analysis.
`
`Basic VIKI
`Hypertext, in its most general sense, allows content (text or
`other media) to appear in multiple contexts. In most
`hypertext systems, including the World-Wide Web, this is
`accomplished by providing nodes or pages of information
`whi~h can be accessed from different through different
`navigational links. Experiences with the use of node-and(cid:173)
`link hypertext
`led
`the development of the spatial
`to
`hypertext model instantiated in VIKI [3].
`
`VIKI provides users with the ability to place and move
`information "symbols" in a set of two-dimensional spaces,
`called collections. Symbols can be assigned visual
`properties such as color, shape, and border width using an
`interface similar to the kind associated with a simple
`graphics editor. These visual properties can be used to
`indicate
`something about
`the
`information,
`i.e.,
`red
`rectangles are from a particular source, or a wide border
`indicates a good article.
`
`The symbols point to information "objects", which contain
`the actual content. These objects may be semi-structured -(cid:173)
`an object may contain a set of textual attribute/value pairs.
`Several symbols may point
`to
`the same underlying
`information object, allowing the same object to be placed in
`multiple spatial contexts.
`
`Objects may be given an object "type ," which specifies a set
`of attributes for the object and a set of visual propertie. for
`the symbo ls pointing to that object. As an example, the
`object type "book" might have the expected attributes title ,
`author, publisher, and synopsis, and commonly be
`presented as symbols with red rectangles with thin borders.
`
`Both the set of attributes and the visual propertie · defined
`by the type are defaults that may be modified for individual
`
`74
`
`objects and symbol - of that type. Some objects of type
`"book" could have an additional editor attribute and no
`value for author. The visual appearance of symbols pointing
`to
`the same object may be different, so one symbol
`representing a particular book object might be blue with
`thick border. even though all the other symbols for the same
`object follow the default for objects of type book.
`
`VIKI's two dimensional spaces are called "collections."
`Collections may contain other collections, creating a
`hierarchy. VIKI users may navigate into (or "maximize") a
`collection -- causing that collection to take the full space of
`the VIKI window -- by double-clicking on the border.
`Double-clicking on the border of a currently maximized
`collection causes the collection to go back to its normal size.
`
`To view the contents of information objects, users double(cid:173)
`click on the border of an information symbol. For objects
`whose contents are stored in VIKI, this opens a viewer/
`editor containing the information. Objects may also point to
`information available via the Web. When a user selects such
`an object for viewing, VIKI launches a Web browser
`(currently either NCSA's Mosaic or Netscape's Navigator
`depending on
`the user's preference)
`to display
`the
`information.
`
`When people are performing analyses in VIKI, they collect
`information, and organize it within the space to reflect their
`under ·tandings. A study of common spatial patterns
`generated
`in
`this
`type of analysis
`task
`in other
`computational and non-computational settings indicated that
`a small set of primitive spatial structures were common to
`many of the resulting arrangements [4).
`
`VIKI uses a heuristic spatial parsing algorithm to identify
`structure implicit in the layout of information symbols
`based on these common structures. The four basic structures
`that VIKI recognizes are stacks, list , composites, and heaps
`(shown in Figure 2). Stacks are overlapping objects of a
`uniform type. Lists are horizontally or vertically aligned
`objects of the same type . Composites are repeating patterns
`of objects of differing types aligned in repeating patterns.
`Heaps are overlapping objects of differing types . Thus, the
`characteristics used
`to recognize structure are overlap,
`alignment, and object type.
`
`D
`D
`D
`
`List
`
`Stack
`
`Heap
`
`• D
`
`D
`
`Composite
`
`Figure 2: Diagram of spatial structures recognized in VIKI.
`
`11 of 15
`
`

`

`--~ .... - ......... _ .......
`
`~,..._-~ ....... ,,.., .... .i..--i......., .• , ....
`
`,,
`; .._..__
`~u.:':="~::::~---.-.- l
`
`.,..~ ........ ~~...-­
`t&Li,.,._ .... .........._.. .. .,~ .......
`
`Figure 3: Black dots show the selection extension resulting from four subsequent mouse clicks on the structure from Figure 1.
`
`By reapplying the basic recognition algorithms to the
`recognized structure, a hierarchic parse of the spatial layout
`is generated. For example, VIKI can recognize lists of
`compo ites which themselves contain lists, such as the
`structure on the right . ide of Figure I.
`
`VIKI uses this recognition in two ways: to support access to
`the implicit spatial structures, and to aid in the use of the
`VIKI data model. Access
`to
`recognized structures
`i ·
`provided through hierarchic click-selection, which is the
`spatial equivalent to the selection extension found in most
`text editors and word processors. Each click on an
`information symbol causes the selection to be extended to
`the next higher level in the spatial parse tree.
`
`Figure 3 shows a series of selections resulting from four
`subsequent clicks on a symbol within the structure on the
`right in Figure l. The structures in this example are lists of
`sections that define a complete path. Each section is itself a
`list of objects, each referring to a Web page. Thus the first
`click . elects a single object; the second click selects the list
`of objects in that segment of the path; the third click selects
`the section (including the label); and the fourth click selects
`the entire path.
`
`tructure recognition to suggest
`VIKI also uses the spatial
`new collections and composites, to help manage space, and
`to provide graphical macros for the creation of common
`symbol layouts. This assistance is only provided upon the
`user's request.
`
`Enhancements for Presentation Authoring
`As the experiences described earlier indicate, this basic
`version of VIKI, with the features just described, was a
`useful tool for the initial stages of path authoring. But the
`experience also indicated that VIKI i ould be enhanced to
`better support the proces , .
`•
`
`Path authoring invariably relied on the use of Web-based
`search engines to find the appropriate Web pages and
`ites.
`Path authors used their Web browser to search and browse
`the Web and copied the URLs of desired pages from the
`Web browser into VIKI. Af