CoNus: Workspace-Based Intuitive Collaboration in Virtual Enterprises
M. Berger, H. Hohl, A. Jarczyk, B. Otto, M. Schneider, G. Völksen
Siemens AG,
Corporate Technology, ZT IK
D - 81730 Munich, Germany
Phone: ++49 89 636 51035
Fax: ++49 89 636 49802
E-mail: michael.berger@mchp.siemens.de
Abstract This paper presents the architecture and user interface
of a generic framework supporting collaboration in virtual enterprises
(see [1]). It is intended to provide users with an
intuitive interface methodology which is independent from the underlying
CSCW components like application sharing, audio/video and file transfer.
The user interface relies on a room metaphor with simple drag & drop
or clicking actions for carrying out functions like conference initiation,
invitation, file transfer or application sharing.
The architecture provides a replicated data model to give users awareness
of any modification in the virtual enterprise. The system supports synchronous,
asynchronous, and autonomous collaboration. A field study shows the usefulness
of the described CSCW framework.
Introduction and Motivation
Collaboration has become highly important to improve production and management
processes in virtual enterprises. The utilization of CSCW or Groupware
systems requires users to change their work strategies. Among other factors,
the acceptance of CSCW systems depends strongly on the usability and flexibility
of the system and on the ability to use applications people are already
familiar with. Groupwork features must be extensions augmenting traditional
single user work. An easy transition between single user work and groupwork
in the virtual enterprise is essential.
The groupwork mode must cover a wide area of cooperation techniques
as synchronous/asynchronous or formal/informal interaction, various group
sizes and different group coordination methodologies. To ease the augmentation
to groupwork utilization, the human computer interface is one central criterion
by which users decide pro or against a CSCW system. Hence, the success
of the virtual enterprise depends strongly on the interaction methodologies
between the human and the computer.
This paper presents a CSCW framework called CoNus (Cooperative Networking
for Groups) which follows the above mentioned requirements. The framework
is open to several CSCW products to be integrated as basic components for
functionalities like file transfer, audio/video conferencing or application
sharing. Currently popular CSCW products imply mostly a conference model
and have impact on criteria like group size, networks and network protocols,
and system platforms. The CoNus framework is a generic system adaptable
to several CSCW services. The only request to such components is to offer
a sufficient API in order to be started, cancelled and controlled by the
CoNus framework. Since the application sharing component and the audio/video
system may be independent from each other and definitely from the framework,
CoNus provides a conference model additionally.
Beside the flexibility of a generic system, the user interface for
groupwork is the other main focus of CoNus. Users will prefer to continue
their work with single user applications they are familiar with and application
sharing guarantees utilization of these applications in a group context.
Users will not have to learn a new interaction methodology when working
with the application. However, controlling a group context needs new interactions
and functionalities beyond individual work methodologies. Particularly,
the human computer interaction will have to provide certain entities and
related functions to control human-human interaction. To fullfil this requirements,
a new interaction metaphor is needed which should offer intuitive and simple
usage to ease the step from computer supported individual work to computer
supported collaboration. CoNus uses the room metaphor where each room visualizes
a workspace of an individual person or a defined group. A floorplan displays
an overview on users currently connected to the system and the workspace
they are in. The virtual enterprise is represented by one floorplan. Navigation
from room to room implies context switches from workspace to workspace
inherently starting or terminating conferencing components. User may therefore
switch from their original company workspace into that of the temporarily
existing virtual company workspace.
Beside meaningful entities representing groupwork, the actions for
applying these entities must be simple and intuitive. CoNus is restricted
to drag & drop gestures and clicking, leaving aside long lasting searches
in complexly structured pop-up or pull-down menus, or selection boxes.
CoNus was designed for small groups of coworkers who join intendedly or
occasionally without leaving their desk to move into a specially equipped
meeting room. This fits perfectly into the intention of working part-time
in a virtual company without an overhead of moving into other buildings.
The following chapter gives an overview the CoNus interface including
a sample session. Next, the design and architecture of the CoNus framework
will be characterized and the scenario and result of a field study with
a technical writer and a translator will be presented. Finally, conclusion
and prospects on future work are given.
The CoNus User Interface
Compared to single user applications, cooperative systems require an extended
user interface for group support. The desk-top metaphor appears to be not
suitable anymore with respect to groupwork. While the desk-top is a single
user equipment and the desk-top metaphor is excellent to describe individual
work, a new metaphor is necessary to express cooperation. The room metaphor
is an excellent representative for spontaneously initialized or planned
meetings: a room is both an individual and a group equipment which people
enter or leave, where they communicate or develop documents, and where
they leave notes or retrieve information. Furthermore, a room can be locked
in case its owner is not present or does not want to be disturbed. In particular,
rooms may have several owners - or the owner is a group of users - and
consequently all data and objects are accessible for all group members.
Each room visualizes a specific workspace which is characteristic for its
user.
In the following, the user interface will be introduced in detail succeeded
by the description of a sample session which applies a representative subset
of the CoNus group functions.
The User Interface Objects, Actions and Functions
Objects of the CoNus user interface include rooms, floorplans, walls, overhead
projectors, doors, briefcases, and mail boxes.
Rooms The main object of the CoNus user interface is the room.
Rooms are basic structural and organizational units which are one-to-one
related to individual users as private workspaces or to projects or organizational
units as shared workspaces. Each room represents a certain workspace with
a specific work context and provides all users sharing a room with an equivalent
view according to the WYSIWIS principle.
Floorplans To receive an overview on users and their current
location, a floorplan displays a set of room symbols with icons representing
users in the virtual enterprise. The neighborhood of rooms in the floorplan
does not necessarily reflect the geographical distance between the locations
of the users or their machines. Since the floorplan may be organized individually,
users can place rooms of frequently contacted co-workers close to their's
while placing others into the periphery. Obviously, the floorplan shows
a section of a corporate model. It can be extended by an address book in
order to get in touch with users not represented by room symbols in the
floorplan. One special room is the hallway which is surrounded by all other
room symbols. It is a place for accidental meeting, for public information,
and for navigating to other rooms (see below). As one conclusion, CoNus
visualizes rooms twofold: first as a symbol in the floorplan and second
as a workspace.
Walls The room contains some equipment of which the walls are
of basic meaning. They can be used as huge pin boards and hence document
and folder objects may be placed or moved on them. Consequently, the walls
of one room are closely related to a specific directory in the file system
of the underlying operating system. Document objects on the walls contain
a reference to a particular file in the related directory and a second
reference to the application used to modify the file. Folder objects represent
subdirectories and may contain documents of further folders.
Overhead Projector Each room contains a table with an overhead
projector on top which is the metaphor for starting a shared application.
A document symbol on the overhead projector starts the related application
with the specified file. If a double-click were the gesture to start an
application, this action would not be visible for other group members and
particularly for late comers. In contrast, a document symbol on the overhead
projector tells all group members that an application is running and which
one it is.
Doors Doors are visualized in both the room as the workspace
and the room symbol in the floorplan. The state of a room is open or closed
which allows users to navigate into the related workspace or not. It is
an additional feature of data security on workspace or directory level.
Furthermore, a door is one means to navigate from a room to the hallway
and from there through one of several doors to another room.
Briefcase Finally, the briefcase symbol is an entity related
one-to-one to each user. It is always located in the same room where the
briefcase owner is. The briefcase transports the contained objects to all
the places concurrently with the owner and hence performs a file transfer.
Mailbox One additional window beside the two for the room and
the floorplan contains a mailbox which provides users with the ability
of asynchronous collaboration.
Actions To utilize the CoNus entities described above, the user
must perform two kinds of simple actions or gestures only: a mouse click
for selection and a drag & drop for moving any type of object. The
meaning of these actions relies on the specific object selected and on
the type of objects from where it is dragged and to which it is dropped.
There are basically five functions users can execute by clicking or
drag & dropping. These are navigation, invitation, file transfer, starting
and finishing a shared application, data hiding and message passing.
Navigation To switch from one work context to any other, users
can navigate. Navigation may be accompanied by a switch from single user
mode to a collaborative work mode in case the target workspace is already
populated and the door of the related room is open. A user executes a navigation
by dragging his or her own icon from the room in the floorplan he or she
is currently located in and drops it into the desired room. Another way
of navigating is - as mentioned above - to click on a door. The result
is also a context switch into the workspace of the room related to the
door.
Invitation Beside actively navigating to other rooms, users
may invite others to a certain room. The gesture is the same as the one
for navigation: The icon of the selected person must be dragged to the
room in the floorplan and dropped into the target room where a meeting
should take place. Note that this gesture is really an invitation rather
than an order. The invited person receives a message window which notifies
of the invitation, the invitor and the target room and leaves the decision
to accept the invitation to the user. The invitor only receives a confirmation
that the invitation was delivered.
File Transfer To move a document to the workspace a user navigates
to, he or she must drag the document from the wall and drop it into the
briefcase. After the navigation is finished, the document is located in
the new work context and may be dragged and dropped onto a wall.
Document Sharing To jointly edit a document, the document symbol
has to be dropped onto the overhead projector. The application referred
to by the document symbol is started with the specified file and shared
with all users currently in the room. This gesture should also be used
in single user mode since in case another person navigates into the room
the application is immediately shared to the visitor, as well. Sharing
an application can be finished by dragging the related document from the
overhead projector and dropping it somewhere else.
Data Hiding To avoid data misuse in case a person is absent
from his or her own room, the room may be locked by clicking on the door
of the room in the floorplan. A second click opens the room again. Only
the owner of a room is allowed to open or lock it. A locked door avoids
users from accessing the related room. Additionally, document symbols can
only be dragged from or dropped into a briefcase by the owner of it.
Messaging One small feature presents asynchronous collaboration
by generating message symbols and filling them with a text. The message
object is delivered to another user by dragging it from the wall and dropping
it into the room in the floorplan which belongs to the receiver.
Potential of the CoNus User Interface The user interface currently
guarantees awareness about all conferences for all users. This helps to
see what is going on in the subset of the organization visualized in the
floorplan. In near future, another feature that allows also hiding of users
will be integrated in order to give people means to keep privacy. The consistently
employed WYSIWIS principle creates a feeling of team identity, common goals
and trust.
A Sample Session
Imagine two engineers User_1 and User_2 from two different car manufacturers
designing a common component for their cars. They work at remote places
but they both use CoNus as the framework for their virtual enterprise with
its floorplan and room (see Figure 1).
Figure 1: User_1 at Site 1
User_2 works in a different context with a document on the overhead
projector. The application related to the document is displayed on the
right of the screen. To make a design decision, he needs support by User_1
and so he invites him to his room by dragging User_1's icon from room in
the floorplan and dropping it into his own room (see Figure 2).
Figure 2: User_2 at Site 2 with an application on the overhead
projector
In return, User_2 receives a confirmation that his invitation was delivered
to User_1 but nothing else (see Figure 3).
Figure 3: User_2 invites User_1 and receives a conformation
User_1 receives the invitation in a window including an image of the
invitor. User_1 clicks "OK" on the invitation window, puts a relevant document
into his briefcase by performing a drag & drop gesture, locks his room
by clicking on the door of his room in the floorplan, and navigates to
User_2 by dragging his own icon from his room in the floorplan and dropping
it into the room of User_2 (see Figure 4).
Figure 4: User_1 accepts the invitation
When the drag & drop gesture is finished, User_1 observes the context
switch to User_2: The briefcase with the document appears in the room of
User_2 which actually indicates that a file transfer was executed. The
icon of User_1 appears in the room of User_2 in the floorplan which displays
to all other users that a conference was started in User_2's room. Next,
the audio/video connection is installed such that both users can see and
welcome each other. The application started by User_2 is automatically
shared to User_1, as well. Each user receives a cursor for pointing and
gestures. A token manager shows which one of the users may produce input
to the application and hence which cursor is also for input and which one
for gestures only. A token switch is performed by clicking on the button
in the token manager window (see Figure 5).
Figure 5: User_1 after the context switch to User_2; they share
the same view
The CoNus conference model does not provide a token request queue, a
chair person or any other methodology to manage token requests. This coordination
is left to the users and their social behavior and skills. After User_1
and User_2 finished their collaboration, User_1 returns to his room, navigating
via the hallway by clicking doors instead of dragging and dropping his
icon (see Figure 5).
CoNus System Design and Architecture
The CoNus system relies on a client-server architecture with the server
for communication rather than for centralized data management. Hence, CoNus
specific data are replicated on each client's site and all actions performed
in the any CoNus workspace are wrapped into messages and sent to the server
which multicasts them to all other sites including the one that generated
the action. Each client interprets incoming messages and executes them
accordingly.
The Information Layer
Each CoNus client is constructed on a layered system architecture with
well-defined interfaces to acquainted modules (see Figure 6). The interface
to the user was already described in detail. Beneath the user interface
module, the CoNus Information Layer provides objects as those for rooms,
walls, documents, overhead projectors, briefcases, etc. The objects are
linked according to their relations. For example, a document is placed
on a wall, a briefcase is located in a room, or a user is owner of a room,
etc. The CoNus framework is implemented in Smalltalk and the Smalltalk
Model-View-Controller concept allows programmers to define e.g. one model
object Room with two view objects, one in the floorplan and one as the
workspace. Actions on different view objects of the same model can easily
be interpreted in their specific context: A click on a door of a room in
the floorplan means opening or locking the door while clicking on the door
in the workspace means navigation to the room next to the door. Model objects
are part of the Information Layer while view and control objects belong
to the User Interface Layer.
Figure 6: CoNus Client Architecture
Basic operations in the CoNus object network are more or less graph
operations: generate or remove an object, generate or remove a link and
change the value of an object attribute. Additionally, there are some composed
operations like "move an object" or "change a link target object". Parentheses
guarantee a sequence of operations to be executed as one logical operation.
The Communication Layer
The Communication Component includes the CoNus conference model and the
link to the communication server. The server guarantees synchronized message
multicast to all clients. This strategy implies consistency of replicated
data on all sites. Furthermore, disconnection of clients is possible in
case of high network load, a network breakdown or an intended disconnection
by a user. The server and the disconnected client each set up a history
of executed operations which will be used in a re-coupling phase for conflict
detection and consistency generation (see [3]).
The Middleware Layer
The use of a middleware layer enables the transparent connection to
various present and future network platforms. The Communication Layer relies
on the middleware functionalities and its specific interface protocols
which can be adapted to other middleware easily.
CSCW Components
Out-of-the-shelf CSCW components for application sharing and audio/video
are not part of but utilized by the CoNus framework. Both components are
addressed through the Conference Manager which translates CoNus functions
like navigation or starting a shared application into the API protocols
of the connected CSCW components.
Current CoNus Implementations
There are currently two versions on CoNus: one based on Unix workstations
and one on PCs. The Unix version provides separate components for audio/video
(Parallax) and application sharing (GroupX). The PC version is build upon
the ProShare Video System 200 for audio/videoconferencing and application
Sharing via ISDN.
Further characteristics of the Unix version are: Sun Sparc workstations
with SunOS 4.1.3, OpenWindows, and NFS. The connections runs on an Ethernet
with Socket connections for CoNus clients and RPC calls for controlling
the application sharing and A/V components. The implementation language
is Smalltalk VisualWorks 2.0.
The characteristics of the PC version are: Pentium PCs with Intel ProShare
Video System HW including one card for Audio/video and one for ISDN),SW
and the ProShare Developers Kit. The software platform includes Windows
3.11, DLL and C Connect for linking ProShare Manager via a C interface
with Smalltalk VisualWorks 2.0, WinSock and TCP/IP interface.
The implementation over ISDN had an impact on the conference model.
A navigation between rooms implies the installation of a connection which
will be unsuccessful in case the co-worker is already in another conference.
The awareness of locations of users has been gone since no continuous connection
among the CoNus clients is possible unless all clients are permanently
connected via ISDN to an MCU. The later case normally isn't practiced because
every connection costs during connection time. So the normal case is that
a connection is only established during conference time.
CoNus Test Experience
The CoNus System was tested in a representative real working environment.
The aim of the CoNus field study was to determine if synchronous cooperation
via a tool like CoNus can improve the cooperation via fax, e-mail, meetings
and phone contacts significantly, and if the CoNus system is the suitable
tool to do it.
Although the CoNus field study was not an empirical general test but
a test for a specific scenario and environment with special aims, the test
results allow an extraction of some general conclusions about CoNus features
that are valid also in other domains (see [8]).
Test Environment
In a one week field study with seven test sessions each with a duration
between one and two hours, an editor from a technical documentation company
collaborated with a translator from the a translation center using CoNus
in a point-to-point connection. The video connection was switched off since
both test persons have been familiar to each other well enough. The social
skills like willingness and ability to collaborate was present at both
sites.
Figure 7: CoNus Field Study Scenario
The object of joint work was a technical documentation written in german
to which some paragraphs had to be added and the translation center had
to translate these paragraphs into english. FrameMaker was used as the
source and target documentation system for writing the manuals.
CoNus team members observed six test sessions, one observer at each
site. Furthermore, the test persons filled questionnaires and discussed
their opinions on the CoNus with the developer team.
Test Observations
After a short introduction to the CoNus system, the test persons were able
to concentrate on their work without taking attention of the collaboration
tool. One further user unexperienced with both CoNus and FrameMaker was
able to contribute to the translation process after a brief introduction
phase.
The room metaphor gave a perfect awareness of the current situation
and made users think of really working in the same room. The feeling of
working in the same room was intensive to such an extent that sometimes
one test persons pointed onto the screen using her finger instead of the
mouse.
During the CoNus sessions, the translator asked questions about the
contents of the german text which was not fully clear. Together they improved
the original german text and developed a clear english translation. Furthermore,
the editor and the translator spontaneously exchanged some experience of
using FrameMaker.
The test persons worked always on two documents: the original german
text and the english translation. Sometimes, the window of the english
document was on top at one site while at the other site the german text
was on top which lead to some minor confusion since none of the users did
recognize immediately when the other edited the document. As a consequence,
the test persons desired a strict WYSIWIS on the whole screen beside the
CoNus workspace and the shared windows. Furthermore, the test persons recommended
that the token manager windows should always be on top in order to change
the input right quickly. A key combination for requesting the token could
also be useful in case the token manager window is hidden under a stack
of other windows.
Test Result Summary
One of the most important results of the field study was that the CoNus
system is very easy to learn and to use. Additionally, it does not disturb
collaboration by drawing the attention from the object of common interest.
Furthermore, it offers a high level of conference awareness. Working with
a tool like CoNus has impact on the cooperation style. It gives new chances
of working efficiently and effectively together and may - as a side effect
- also improve the quality of work and the use of shared tools like in
this case FrameMaker.
Related Work
CoNus rooms and their intuitive usage provide users with two basic functionalities:
first, rooms are a meeting facility with features for A/V and data sharing,
and second, they are - in case of project workspaces - shared repositories.
The idea of the room metaphor for collaborative workspaces is not new.
The Xerox Rooms System [6] used rooms to organize the
data of individual users and hence pronounce the repository character.
Other systems using the room metaphor are DIVA [10]
and Argo [5] which employ shared workspaces as meeting
facilities including coffee rooms or hallways to run into coworkers occasionally.
In addition, DIVA supports the repository function of rooms but users have
no opportunity to look their private rooms for reasons of data security
as suggested in [2], or to disconnect for autonomous
collaboration phases.
CoNus does not show users both an individual and a shared workspace
at the same time as presented e.g. in [7]. Since each
room may serve as a meeting facility, users should meet where relevant
common data are located and the briefcase enables them to bring relevant
documents from individual workspaces. The same holds for the DOLPHIN system
[11] which additionally offers LiveBoard equipped
rooms for larger groups and planned meetings.
Ideas presented in the WORLDS system [12] are
also close to CoNus except the awareness of coworkers in the segment of
the organization a user organized and views in the floorplan.
The DIVE system (see [4] and [2])and
the CAD system described in [9] use a 3-D visualization
to display the shared workspaces. This enables users to break the WYSIWIS
paradigm making collaboration supported by audio more difficult. CoNus
uses a 2.5-D visualization which provides sufficient simplicity. Furthermore,
CoNus is not intended to present virtual reality but to show an abstraction
of the reality with no information overflow for the users in contrast to
3-D multi user games.
Compared to CoNus, the systems mentioned here do not provide a generic
framework for supporting different services, applications, and audio/video
components which users may control by using the same gestures and actions.
Conclusions and Prospects
CoNus is currently a demonstrator. A wide range of possible CoNus scenarios
include distributed engineering, tele-learning, tele-medicine, technical
support, hotline, and others which add value to production, products, and
management of enterprises.
In the long term, integration of agent cooperation for carrying out
routine work is in the focus development (see [13]).
A study is under development which evaluates the usefulness of CoNus and
other CSCW systems on personal digital agents. The integration of a multimedia
mail system will be attacked soon. Platform interoperability is another
aim of the CoNus development. This will be garantied by observing the conferencing
standards T.120 and H.320 / H.323 in the conference-management and by using
conferencing tools that are conform to these standards.
References
[1] Byrne, J. A.:
"The Virtual Corporation",
in: International Business Week, Feb 8, 1993
[2] Bnford,
S., Bowers, J., Fahlén, Lennart, Greenhalgh, C. & Snowdon, D.:
"User Embodiment in Collaborative Virtual Environments",
in: Proceedings of the CHI `95 Conference, Denver, CO, May 5-11, 1995
[3] Berger, M.:
"CoNus - A CSCW System Supporting Synchronous, Asynchronous and Autonomous
Collaboration",
in: "Proceedings of the International Conference on Distributed Platforms
/ Industrial Stream",
Dresden, Feb 27 - March 1, 1996
[4] Fahlen,
L.E., Brown,C.G., Stahl. O. & Carlsson, C.A.:
"A Space Based Model for User Interaction in Shared Synthetic Environments",
in: Proceedings of the CHI `93 Conference, Amsterdam, The Netherlands,
Apr 24-29, 1993
[5] Gajewska,
H., Manasse, M. & Redell, D.:
"Argohall: Adding Support for Group Awareness to the Argo Telecollaboration
System"
in: UIST `95 Conference, Pittsburgh, PA, Nov 14-17, 1995
[6] Henderson,
D.A. & Card S. K.:
"Rooms: The Use of Multiple Virtual Workspaces to Reduce Space Contention
in a Windowbased
Graphical User Interface",
in: ACM Transactions on Graphics 5, 3, July 1986
[7] Ohkubo,
M. & Ishii, H.:
"Design and Implementation of a Shared Workspace by Integrating Individual
Workspaces",
in: Proceedings of the 1990 Conference on Office and Information System",
Cambridge, MA, April 25-27 1990
[8] Otto, B., Schneider
M. & Krämer, G.:
"Computerunterstützung räumlich verteilter Teams durch CoNus"
in: Conference on Technical Documentation, Paderborn, Germany, March 22-23,
1995
[9] Shu, L.
& Flowers, W.:
"Groupware Experiences in Three-Dimensional Computer-Aided Design",
in: Proceedings of the CSCW `92 Conference, Toronto, Canada, Oct 31 - Nov
4, 1992
[10] Sohlenkamp, M. & Chwelos
G.:
"Integrating Communication, Cooperation, and Awareness: The DIVA Virtual
Office Environment",
in: Proceeding of the CSCW `94 Conference, Chapel Hill, NC, Oct 22-26,
1994
[11] Streitz, N., Geißler,
J., Haake, J.M. & Hol, J.:
DOLPHIN: Integrated Support across Local and Remote Desktop Environments
and Liveboards",
in: Proceedings of the CSCW `94 Conference, Chapel Hill, NC, Oct 22-26,
1994
[12] Tolone, W.J., Kaplan, S.M.
& Fitzpatrick, G.:
"Specifying Dynamic Support for Collaborative Work within WORLDS"
in: Proceedings of the COOCS `95, Milpitas, CA, Aug 13-16, 1995
[13] Völksen, G., Haugeneder,
H., Jarczyk, A. & Löffler, P.:
"Cooperation-Ware:
Integration of Human Collaboration with Agent-Based Interaction", in:
"Proceedings of Second International Working Conference on Cooperating
Knowledge-Based
Systems", Dake Centre, University Of Keele, Keele, Staffordshire, UK, June
1994