Software

Collaborative Personal Information Management With Shared, Interactive Tabletops

Description
Given the central and routine nature of Personal Informa-tion Management (PIM) to conducting daily tasks, the abil-ity to conduct PIM with tabletop interfaces (and other col-laborative devices that diverge from traditional desktop in-teraction) is an
Categories
Published
of 10
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Collaborative Personal Information Managementwith Shared, Interactive Tabletops Anthony Collins School of Information TechnologiesUniversity of Sydney, Australiaanthony@it.usyd.edu.au Judy Kay School of Information TechnologiesUniversity of Sydney, Australia judy@it.usyd.edu.au ABSTRACT Given the central and routine nature of Personal Informa-tion Management (PIM) to conducting daily tasks, the abil-ity to conduct PIM with tabletop interfaces (and other col-laborative devices that diverge from traditional desktop in-teraction) is an important method of supporting collabora-tive work at the tabletop. Using tabletops potentially allowspeople to conduct PIM in a comfortable place that is usedfor a range of regular activities, from eating to reading andwriting. However, there are challenges in supporting naturaland useful PIM with the properties and limitations of table-top interaction. The use of large surfaces also provides greatpossibilities for collaboration, as PIM often involves otherpeople, or people may want to selectively share their per-sonal information with others for a variety of purposes (suchas organising work e-mail with colleagues).In this paper, we present our techniques for navigating andsorting multiple sets of personal information—particularlydigital files and e-mail—on a tabletop. We present  Focus ,a multi-user tabletop interface where users select “focusitems” that retrieve related artefacts from unified personalinformation spaces, with minimal prior configuration. Weconclude with a discussion of results obtained in a small-scale user study, where participants were asked to access andorganise personal information at an interactive tabletop. Author Keywords Personal Information Management, Tabletop Interface, Sin-gle Display Groupware ACM Classification Keywords I.3.6 Computer Graphics: Methodology and Techniques— interaction techniques ; D.2.2 Software Engineering: DesignTools and Techniques— user interfaces ; H.5.3 InformationInterfaces and Presentation: Group and Organization Inter-faces Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise, orrepublish, to post on servers or to redistribute to lists, requires prior specificpermission and/or a fee. CHI 2008  , April 5 - 10, 2008, Florence, Italy.Copyright 2008 ACM 1-59593-178-3/07/0004... $ 5.00. INTRODUCTION Personal Information Management (PIM) is a task that istypically performed with a conventional personal computer,by a single user. Moving away from the desktop paradigmand its single-user interaction methods allows new possibil-ities for both sharing personal information, and collaborat-ing on the management process. The interactive tabletop—a novel medium that has recently attracted significant re-search interest—supports collocated collaboration with alarge, shared workspace, where people can sit face-to-faceand interact with it simultaneously.While PIM is typically a personal task, we would like to sup-port people naturally sharing and accessing their personal in-formation at a shared, interactive tabletop. A key facility of a tabletop is to support interaction with digital files for con-ducting collaborative tasks. Considering the central natureof other types of personal information (such as e-mail) toconducting daily tasks, support for accessing and organisingthis information at the tabletop can potentially be very use-ful. For example, an office-worker may want to share work related e-mails and files with a colleague in order to performdecision making. Another example is two people workingon a joint project involving different types of information,where they have divided the work. The pair would like tocombine their work and discuss it using a collaborative in-terface.Despite the advantages of the tabletop medium for collo-cated collaboration, the properties and constraints of table-top interaction call for re-thinking standard approaches toaccessing and managing personal information. First, it mustbe natural to interact with the tabletop using input with spe-cial constraints not present in personal computer interaction:a keyboard and mouse is typically not present in a multi-usertabletop setting. While providing a projected keyboard onthe tabletop surface is possible, this provides no tactile feed-back when pressing keys, and does not give users a fixedreference of where to place their hands.Second, tabletop interfaces must support people sitting in arange of orientations, as a fundamental facility is to supportpeople sitting both face-to-face and around-the-table. Con-sequently, interface elements must be orientation indepen-dent. This makes the use of text (such as file and directorynames) on a tabletop problematic, as it may be difficult toread text from all positions around the tabletop.  Third, the size of the tabletop has an impact on a numberof design factors, such as the size at which to display textand interface elements. More importantly, the limitations of physical human reach and direct-touch interaction must beconsidered, as a user may not be able to reach across thetable. User interface selection targets must also be suffi-ciently large for direct-touch interaction, which potentiallycontributes to clutter on the tabletop. A tabletop PIM inter-face must support users working with a particular set of per-sonal information at any position around the tabletop—thismeans that the point of interaction with a personal informa-tion set must not be dependent on a particular position (ororientation) on the tabletop surface.Finally, when considering tabletop size, display resolution,and multiple users, clutter is a critical problem to addressin tabletop interfaces. Given the increasingly large size of personal information collections, the need to present largeamounts of information, or the need to replicate informationon the tabletop for viewing or interaction by multiple peo-ple in different orientations, is a major problem when de-signing to minimise clutter. Multiple people may create andinteract with artefacts that overlap both shared and individ-ual objects on the tabletop, particularly when artefacts arezoomed-in for a more detailed view. In a multi-user setting,where several people are creating and interacting with con-tent on the tabletop simultaneously, management of clutterbecomes challenging—existing methods to manage clutterin conventional personal computers, such as a ‘task-bar’ ina window manager to control active windows, are only de-signed for single-user interaction. Another challenge is thatthe management of clutter by one user could have uninten-tional and adverse affects on other users of the tabletop.When taking into account the properties of tabletops, andtheir collaborative nature, we need to explore ways to accessand organise digital information in a natural way at the table-top. The unified approach to personal information manage-ment is ideal for conducting tasks at the tabletop, as it movesaway from users navigating separate  silos  of information atthe tabletop (such as uniquely organised hierarchies for filesystems and e-mail). In this paper, we explore multi-user,unified access to e-mail, personal files, photos, and storedweb-pages using interactive tabletops. RELATED WORK A significant problem in modern personal information man-agement is information fragmentation [16], where peopleregularly need to access and manage related information inseparate physical locations with little support from the toolsthey use. Studies conducted by Boardman and Sasse [4]highlighted that people employ a range of personal strate-gies to manage their information, both within and acrossdifferent PIM tools. The notion of   folder overlap  was ob-served for many participants, particularly for files and e-mail, where people create a similar hierarchical organisationbetween the two information spaces. As we are exploringcollaborative PIM using a tabletop, we must support peo-ple concurrently interacting with information that is organ-ised differently, both within a person’s information collec-tion, and between multiple people’s information collections.The  Stuff I’ve Seen (SIS)  interface provides a unified indexto all files and personal information that a user has seen ontheir computer [8]. SIS does not rely on a hierarchical or-ganisation structure, and uses contextual cues (such as datesand people) to facilitate information retrieval. This is sim-ilar to  Lifestreams  [12], and  MyLifeBits  [13], where the in-formation a user has seen is ordered on a timeline for laterretrieval. Improved search facilities for personal informa-tion, which merge searching and browsing, were presentedwith  Phlat   [7]. Unified tagging allows users to create anorganisation scheme consistent across all of their personalinformation. Furthermore, Phlat (like SIS) presents this in-formation in a single interface, thereby removing the needto use separate tools to access different types of information.These projects tackle the problem of supporting a unifiedpersonal space of information.Content-basedsearch, suchaswith GoogleDesktopSearch 1 ,  Apple Spotlight  2 , and  Windows Vista Instant Search 3 , alsoallowsinteractionwithasingleinformation-spacethatmergesrelevant content regardless of its type and where it is stored.However, these content-based access mechanisms provide atargeted search interface that would be difficult to adapt toa collaborative tabletop, where there is a need to support in-teraction with multiple personal collections that may oftencontain related content.  Haystack   [15] is a general-purpose information managementtool that allows a variety of information types to be managedwithin a single user interface. Users can also create arbitraryrelationships between the stored information to simplify re-trieval. Karger  et al.  ( ibid  .) discuss the concept of present-ing “similar items” alongside user-selected items, in order tosupport iterative query refinement, which is similar to ideasexplored in the  IQ prototype  [9], where information relevantto the the user’s current context is presented in a peripheralinterface.The notion of   Group Information Management (GIM)  hasbeen a recent topic of discussion [10]. The focus of GIM ison sharing information—that was created with a clear intentto share—with groups and institutions, rather than individ-uals in a collocated context. Our focus in this research ismore accurately viewed as collaborative PIM (rather thanGIM) where people work collaboratively with their personalinformation, or they selectively share parts of their personalinformation with a small number of people in a collocatedcontext.PIM research has focused on making organisation and re-trieval easier, but there has been little exploration of con-ducting PIM with devices that are inherently collaborative.These devices need to support people sharing their personalinformation with others, as well as helping people collabo-ratively navigate multiple sets of information. 1 http://desktop.google.com/ 2 http://developer.apple.com/macosx/spotlight.html 3 http://www.microsoft.com/vista/  Tabletop and pen-based interface research has explored in-teraction with small collections of information, such as digi-tal photographs [3], or a virtual desktop of files [1]. Pure hi- erarchical interaction has been explored in the  Personal Dig-ital Historian (PDH)  project [23], as a way of increasing thescalability of a tabletop interface. Interaction through pas-sive and active associations has also been explored [22], al-though results suggest that participants highlighted “clutter”and “over crowding” within groups of information as majorproblems in the interface. This crowding issue is noted as animportant factor to address in tabletop interfaces [19].Research into PIM at a tabletop has been limited. Earlierwork on  micro-mobility  [11] has researched people explic-itly and visibly transferring files between devices, such astabletops and personal computers. Other approaches re-quire users to move their information into a shared regionof the tabletop from a laptop or personal storage device [14,21]. These systems do not support people interacting withbroader collections of personal information at a tabletop.Techniques for collaborative searching of large collectionson tabletops have been explored in  TeamSearch  [17], al-though these require manual Boolean query formation.However, the benefits of collaborative searching are high-lighted, suchasfacilitatingstrongercollaborationandaware-ness among group members. Providing a search interfacealone is not sufficient for supporting natural browsing of multiple sets of personal information on a tabletop.We previously explored  OnTop  [6] as an interface for col-laboratively accessing personal file systems from a table-top. Our evaluation showed that OnTop provided valuablesupport for tabletop collaboration on the  access  process—participants were more social and co-operative when usingOnTop instead of a hierarchical alternative (which encour-aged private file system access and selective sharing of filesonce they had been located). As OnTop presents informationaccording to its content, rather than storage location in a filehierarchy, the techniques used are applicable to a variety of organisation schemes and types of personal information. BRINGING PIM TO THE TABLETOP Previous PIM research has explored different techniquesto support people interacting with their personal informa-tion [7, 8, 12, 13, 15, 18]. Support for people interacting with group information is also an active research area [10].However, there has been little research into supporting peo-ple conducting PIM using collaborative, multi-user devices,such as tabletop displays. Despite this, there is a need tosupport PIM at the tabletop in order to facilitate people col-laborating with personal information in a collocated setting.The following scenario illustrates collaborative PIM using atabletop interface:Jim is eating breakfast while sitting at his dining table,whichisalsoaninteractivetabletop. Ashiswife, Sarah,is also sitting at the table, they wish to make somedecisions about an upcoming holiday trip. They havebeen organising important information using e-mail,text documents, photos, and stored web-pages, on eachof their personal computers. They must make somebookings during the day, and they have decided to splitthe workload. Jim needs to access his e-mail (stored onhis desktop computer in another room), which containsquotes for hotel bookings, and needs to exchange thiswith Sarah who is organising transport to and from thehotel. Sarah needs to exchange details about when sheis able to take time off work for the holiday, with the in-formation stored on Sarah’s desktop computer (also inanother room). They use the tabletop to jointly reviewe-mails, files, photos and web-pages, in order to makeimportant decisions about the trip, and to delegate dif-ferent aspects of the planning.This scenario involves sharing of personal information thatneeds to come from multiple remote computers. Due tothe context of use of tabletops (which will often be used inshared, open environments), we must support people inter-acting with PIM collections that are stored remotely—we donot envisage that a tabletop is going to be used as a desk-top computer replacement. Rather, tabletops and other de-vices (such as desktop and laptop computers) must operatetogether harmoniously. Thisrequiresuniversal accesstoper-sonalinformationanditsassociatedmeta-datafromavarietyof devices in a pervasive computing environment. We focuson providing support for collaborative PIM using a tabletop.When supporting multiple people sharing their personal in-formation, we need to facilitate interaction that is indepen-dent of both the PIM tools used and the underlying storagelocations of the information. Different tools (such as an e-mail client, and a file system interface) may organise relatedinformation in different hierarchies, with little integrationbetween them. When accessing this information at the table-top, weneedtopresentaunifiedviewofthisinformationthatis independent of its storage. If people bring their personalinformation to the tabletop, we need to facilitate natural in-teraction with information from a variety of tools—if oneperson is using a certain e-mail client to store their mail ina folder hierarchy, and another person is using a differentclient that uses tagging instead of hierarchical organisation,we need to support both these people easily interacting withthe mail in a consistent way at the tabletop.Furthermore, in the scenario presented, the two sets of in-formation that are accessed at the tabletop contain highly re-lated information. We need to support collaboration with thecombined set of information at the tabletop. As a key goal of our work is to support people collaborating with each other,it is likely that people will be sharing related information,and so we envisage that there may be significant informationoverlap between collections. Therefore, we need to supportpeople interacting with similar collections of information.With these design considerations in mind, we now outlineour techniques for supporting collaborative PIM at a shared,interactive tabletop.  Figure 1. People using Focus to collaboratively access and organisetheir personal information at a tabletop.Figure 2. A selected e-mail (left), and a flipped-over PDF document(right) with its filename written on the back. FOCUS To provide the functionality discussed in the previous sec-tion, and to address the properties and constraints of table-top interfaces, we have created the  Focus  interface for as-sociative access to multiple sets of personal information atan interactive tabletop. Focus (shown in Figure 1) presentsa unified view of personal information, that is independentof the physical storage location of personal information, andthe tools used to manage it. Focus provides a  remote  in-terface to personal information, where users  export   selectedpersonal information to the tabletop for collaboration. User View Focus shows image representations of personal informationthat appear to be placed on top of the tabletop surface. E-mails, PDF documents, photos and web-pages are presentedand manipulated on the tabletop in an identical way. Theappearance of e-mail on the tabletop is similar to a printede-mail, with a brief header describing the mail recipients andsubject. An artefact can be  moved   by selecting it in the mainarea inside the solid selection lines (shown in Figure 2) anddragging it. When moving objects, they have realistic mo-mentum so they can be  flicked   around the tabletop, makingit easy to move objects to areas of the tabletop that are outof physical reach. Objects on the tabletop can be rotated andresized (in a combined  rosize  gesture) by selecting the objectat one of its corners and dragging the corner.A personal information object can be  flipped   by selecting itfrom within a stippled triangle along one of the its edges and Figure 3. The Focus start view, which shows the contents of a userspecified mailbox in a radial layout on the tabletop. dragging it across to the opposite edge. Once flipped, theuser sees a description of the object written on the back, asshown in Figure 2. For example, if the object is a file, thenthe filename is written on the back.When Focus is first launched, a broad  start view  of the per-sonal information is presented in a radial layout on the table-top (shown in Figure 3). Our current implementation hastwo types of start views available. The first shows the firstfile (alphabetically) in each exported directory of each re-motefilesystem, andthemostrecente-mailineachexportedmailbox. This should work well if users have a reasonableorganisationoftheirfiles, andifthefirstfileinadirectory(orthe most recent e-mail in a mailbox) is representative of thefolder or mailbox content. The second technique presentsonly the contents of a nominated mailbox on the tabletop (inthe same radial layout). From these start views, the contentpresented on the tabletop can be used to navigate to other re-lated information contained in the personal information sets.In this paper, we present and evaluate the mailbox start view.Navigation of personal information is based on the notionof a  focus  item. Once a user selects a focus item, all other related   pieces of information are displayed on the tabletop(regardless of where they are stored or which computer sys-tem and user they belong to). The personal information frommultiple people is merged to appear as a single, combinedinformation set on the tabletop, and a user may navigate theinformation by re-selecting focus items. To select a focusitem, a user dwells (depresses the pen for one second) onit. A ‘click’ sound gives feedback that it has been selected.Matching results from each remote computer are returnedand displayed immediately on the tabletop. This approachallows people to easily find related information in both fa-miliar and unfamiliar collections.Figure 4 shows Focus after two focus item selections fromthe start view. To address the limited display area of thetabletop, personal information is presented in a non-uniformmanner, where the initial size of each items’ image repre-sentation is determined by its relevance to the focus item.Thus, most relevant items appear large and prominent, while  Figure 4. The Focus interface showing a collection of e-mails, photos,PDF documents and stored web-pages, that have been rearranged byusers after two focus item selections. less relevant items appear small and unobtrusive. The resizegesture can be used to override the initial size based on sim-ilarity. If an item was previously displayed on the tabletop,it is presented in its last location if the user moved it to aspecific position. This is to provide spatial consistency be-tween focus selections, and to allow users to create arbitraryspatial groupings of their personal information (as in  Data Mountain  [18]).To reduce clutter on the tabletop, users can move unwanteditems in the  Black Hole  (shown in the bottom right of Fig-ures 3 and 4), a special object that is always visible and above other objects. It can be moved, rotated, and resizedlike any other object, although an object reduces in size asit is dragged closer to the centre of the Black Hole until itis completely hidden. Objects placed in the Black Hole maybe retrieved by reducing the hole’s size and flicking it withsufficient momentum, causing the contained objects to “fallout”.The  History Browser   widget, shown at the top of Figures 3and 4, supports ‘back’ and ‘forward’ navigation operations.This object can be manipulated like any other interface ob- ject, although it cannot be placed in the Black Hole. It showsthumbnail representations of recent focus items (typicallythe last five focus items, depending on the aspect ratio of the thumbnails), and a user dwells on a past focus item togo ‘back’ to the display of information associated with it. Aspecial circular icon on the far left of the History Browserrepresents the srcinal start view, which can be returned toby dwelling on the icon. Storage Bins , shown in Figure 5, are special folder-like ob- jects on the tabletop that show thumbnails of items draggedinto them (similar to [20]). We use a space-filling thumbnaillayout to accommodate the thumbnails placed in the StorageBin. Storage Bins can be moved, rotated and resized on thetabletop just like any other interface object, and they can alsobe placed in the Black Hole. Figure 5. An empty Storage Bin (left) with a pre-defined label, and aStorage Bin containing four items (right) with its label obscured. Itemscan be removed from the Storage Bin by dragging them out. Exporting Personal Information The Focus interface supports people interacting with per-sonal information stored on a remote computer. As a result,Focus is designed so that people export a subset of their per-sonal information to the tabletop when they need to use itin a collaborative setting. To make personal information ac-cessible by the Focus interface, users run the  Focus Exporter  on their personal computer. After a focus selection from thetabletop, a query is sent to each connected machine to findsimilar personal information. This query request containsmeta-data associated with the focus, so that similar informa-tion can be found in separate personal information spacesthat do not contain the srcinal object.A critical issue in supporting this functionality at a collabo-rative tabletop is privacy. To address this, users are given ex-plicit control over the information accessible by tabletops—users may specify (using the Focus Exporter application)which tabletops may connect to the computer to requestsimilar items, and configure which information can be re-turned. Currently, the tabletop access control is based onIP addresses, as it is expected that all the computer systemsare connected to the same local trusted network. Users mayalso configure the start view of their information. This in-volves either selecting a file system path where the first filein each directory, and the most recent e-mail in each mail-box is presented for the start view, or selecting a mailbox tobe displayed on the tabletop. As the start view is configuredon each remote machine, each person can have their owncustomised view presented at the tabletop. Similarity of Personal Information AsnavigationwithFocusisperformedbyassociation, wheresimilar information is presented after a focus item is se-lected, it is essential that Focus will correctly retrieve therelevant information. It would be unreasonable to imple-ment a static relevance calculation mechanism, as peopleuse a variety of tools to mange their personal information,and they employ a variety of different organisation strategiesboth within and across different tools [4]. Accordingly, wehave created a mechanism for users to customise how theirinformation is determined relevant to a given focus—usersselect (on a scale of 0 to 10) how important certain meta-data attributes are to the organisation of their information,as shown in Figure 6. For example, if the user deems infor-mation by the same author(s) (such as the document author,or the sender of an e-mail) as closely related, the user in-creases the weighting on the ‘authors’ attribute. A ‘tool-tip’is displayed when moving the mouse over each meta-dataattribute to describe its meaning. For example, moving the
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks