You are on page 1of 6

Proceedings of the 2009 13th International Conference on Computer Supported Cooperative Work in Design

Supporting Collaborative Virtual Meetings using Multi-Agent Systems


Phil Thompson, Rahat Iqbal, Anne James Distributed System and Modeling Research Group, Coventry University, UK {p.thompson,r.iqbal, a.james}@coventry.ac.uk Abstract
In an increasingly sophisticated global business environment there is often a need to be able to bring together professionals from all types of organizations from all over the world. Those professionals operate under severe time constraints and a conventional faceto-face meeting is both impractical and expensive. Virtual meetings have proved to be a convenient alternative, offering a flexible meeting domain more suitable for the busy executive. This paper explores the concept of using a peer-to-peer network and multiagent systems to facilitate virtual meetings, managing the artifacts required for the session. By the use of a laptop, PDA or similar electronic device the meeting participant can join the meeting and collaborate in the discussion. in time wasted by traveling between locations. The level of presence required at each of the meetings may be different. Those contributions may involve attendance for short or long periods of time. Short contributions could include initiating, concluding or interrupting meetings to pass in some vital information which will influence the proceedings. Longer contributions requiring the need to take part in discussion or play an active part in facilitating the meeting will require more prolonged attendance. Virtual meetings offer a good alternative to face-toface meetings to counter some of these problems because they remove the need to travel to facilitate, initiate, interrupt, take part in discussion or conclude meetings in distant locations. The use of the Internet also makes it easier to bring together executives from different parts of the world using a common mode of input, the web browser. There are many supporting infrastructures for virtual meetings including telepresence [2]; video-conferencing; web conferencing and teleconferencing by calls over a telephone link or using Skype. These vary enormously in sophistication as well as cost. Telepresence is as close to a real meeting as electronic methods can get allowing visibility to all participants as if they were in the same room or at the same table [3]. Teleconferencing has the disadvantage that the participants cannot see each other and everybody may try to speak at once. Apart from telepresence all the other methods however share certain disadvantages [4]. The meeting needs somebody to act as a facilitator to manage the agenda, to make sure that people speak in turn and that they are all looking at the same documents (e.g. minutes, agenda item etc.). Multiple agent based solutions have been developed which handle diaries to book meetings [5,6]. The use of virtual worlds [7] where participants are represented by avatars, are a more recent phenomenon supporting collaboration for meetings and training. In order to address these issues this paper describes a prototype solution mechanism which enhances the virtual meeting environment and addresses some of the disadvantages mentioned above. Section 2 identifies the objectives and theoretical foundation of this paper, Section 3 gives a functional description of the prototype, Section 4 describes the logical design of the prototype, Section 5 explains the technical design of the prototype

Keywords: Virtual Meetings, Peer-to-Peer, Multiagent systems, Collaboration

1. Introduction
Much of the senior executives normal day is spent going from meeting to meeting sometimes within the organization where they work but also involving travel to other locations on a national, international or global scale. Meetings occupy a good proportion of the available time [1] and there is a need to keep the number of meetings to a minimum, make efficient use of the time available and reduce the need to travel. While time is an important factor requiring choices to be made on what meetings are attended according to the priorities and needs of the organisation the amount of money spent is also critical. Considering the average salary of high meeting attenders and how many hours they spend in meetings each year, a lot of money can be wasted if either the time spent is not efficiently used or bad choices are made on which meetings to attend. If additional expenses including travel and subsistence are added to this it could be quite a substantial amount of money wasted. Proper management of the whole process is therefore required. Face-to-face meetings make the whole process complicated for the manager needs to make contributions to meetings in different locations resulting
978-1-4244-3535-7/09/$25.00 2009 IEEE

276

and Section 6 provides conclusions and outlines our future work.

2. Objectives and Theoretical Foundation


The objective of this paper is to investigate into and draw from the theoretical underpinning of multidisciplinary research involving Multi-Agent System and Peer-to-Peer architecture. Based on our investigation, we have developed an efficient and robust system to support collaborative virtual meeting using multi-agent systems in peer-to-peer environment. The meeting agenda and on going discussion is structured using the concepts of IBIS (Issue Based Information System). This section is devoted to provide a brief introduction to Multi-agent systems, Peer-to-Peer and IBIS.

about its environment and its capabilities, desire represents the state of affair the agent want to achieve and intention corresponds to the desires the agent is committed to achieve. BDI architecture is also called deliberative architecture because a BDI based agent is involved in a deliberation process before deciding what actions to take to achieve a particular goal in a given situation [10, 11].

2.1.1

Communication

Language

and

Protocols
The agents communicate using standard Agent Communication Languages FIPA ACL (Agent Communication Language) and KQML (Knowledge Query and Manipulation Language) [16, 17]. We intend to use FIPA ACL for agents communication in the meeting system. The FIPA ACL is built upon speech act theory [18]. The agents standard communication languages facilitate the interaction between independently developed agents belonging to different meeting participants and the meeting facilitator. The interaction protocols defined as a standard sequence of messages that agents follow during interactions. The messages passing between the agents contain information about interaction protocols and the type of the message.

2.1 Multi-Agent systems


In meetings, multi-agent systems support humanhuman communication and collaboration by assisting meeting participants. Agents can perceive its environment through sensors and performs actions on the environment through effectors. Artificial intelligent agents consist of a number of autonomous interacting software entities called agents, whose behaviour is governed by social norms or interaction protocols. In general a Multi-agent System (MAS) is a computational system in which agents with different capabilities and resources perform their task by coordinating and cooperating with each other in order to achieve a set of goals. Michael Wooldridge and Nicholas Jennings [8] argue that the term agent is used to denote a hardware or (more usually) software-based system that should have the following properties: Autonomy: Agents operate without the direct intervention of human or others, and have some kind of control over their actions and internal state. Social Ability: Agents interact with other agents (and possibly humans) via some kind of agent communication language. Reactivity: The agents perceive their environment, and respond in a timely fashion to changes that occur in it.

2.1.2 Coordination
Coordination is a process by which an agent reasons about its local actions and the (anticipated) actions of others to try to ensure that community acts in a coherent manner [18]. The Intelligent agent paradigm provides an effective way of managing flexible coordination at service level using higher level protocols and coordination models. For successful coordination agents require to maintain the model of each others future interactions such as managing the meeting agenda and the ability of the facilitator to handle the next participants request. They use these models in order to guide their future actions.

Pro-activeness: Agents do not simply act in response to their environment; they are also able to exhibit goal-directed behaviour by taking the initiative. Agents are built based on an approximation of the human reasoning process encapsulated in BDI architecture. The BDI architecture of an agent is based on the Bratman theory of practical reasoning in humans [9]. It represents three major components of an agent; known as belief, desire and intention. The belief component represents the information the agent has

2.2 Peer-to-peer
Peer-to-Peer is used to provide a collaborative network platform for the meeting environment. The scalability offered by Peer-to-Peer networks means that it is easy for new participants to join the meeting. In a conventional client-server network extra resources have

277

to be provided at the server as more and more nodes join the network. In a peer-to-peer network [12] no central server is required each peer brings enough extra resource to support its own membership of the network. Peer-toPeer computing also allows shared information to be, accessible by other peers directly, without passing intermediary entities. When a new member joins the team information can be sourced from any of the other members. Although this creates data redundancy it does offer data resilience by providing an answer to the single point of failure problem encountered in central server systems because no central repository is used for information. Figure 1. Participant Screen

3. Functional Analysis
What is required for the virtual meeting scenarios described above is a managed environment. For efficient management of the meeting [13] there needs to be a facilitator. The facilitator needs to create an agenda where the agenda item sequence is clear for all participants and an estimated time for discussion for each item should be added. Ideally somebody should be asked to lead the discussion on each agenda item. At the end of each agenda item the facilitator should summarise findings and then clearly identify what actions are required and who is to perform the action [14]. Ideally a date should be associated with action to ensure feedback. The participants on joining the meeting should be given a copy of the agenda and any other supporting documents. The facilitator should control who speaks and when they speak and should be able to interrupt if required to seek clarification. Participants should be able to signal to the facilitator to gain access to the meeting; when they wish to speak; if they do not understand; if they want to move progress; if they wish to leave the meeting, etc. They should be able to terminate a session and resume at a later time in case they do not need to be present for certain agenda items. At the end of the meeting they should be provided with a copy of the action list by the facilitator. When the participant (see Figure 1) joins the meeting they log on to the meeting environment with the PDA and then send a login message to the facilitator. The facilitator agent intercepts the login message on the way to the facilitator and sends a list of attendees to the participant as shown in the example. On receiving the login message the facilitator then sends a copy of the agenda to the participant which will be stored in the PDA and the current agenda item updated with the first agenda item description. Any other supporting documents (artifacts) will also be sent to the participant, stored on the PDA and can be displayed on the screen if required. Messages to alert the facilitator (request to speak, clarify, etc.) will be sent from the PDA by the participant which will be intercepted by the facilitators computer and displayed on the screen. The participant will be able to see whether the messages have been responded to by the facilitator and their status.

4. Prototype Design
We designed a prototype environment (diagrams and descriptions developed from [15] operating in its own window to be used exclusively if accompanying a teleconference or to be used in combination with windows used to support web-conferencing or videoconferencing. The environment is capable of operating on portable devices like laptops or PDAs which support a windows type operating system.

Figure 2. Facilitator Screen The facilitators computer screen (See figure 2) displays messages from the participants as well as allowing messages, agenda, and any other artifacts to be sent to them. The facilitator selects participants and then sends messages to control the flow of the meeting (e.g.

278

speak now, stop speaking, clarify point, etc.). As the meeting progresses the facilitator terminates an agenda item and adds actions, participant names and dates to the agenda database. On terminating agenda items a message is sent to the participants in order to update the current agenda item with the next agenda item description item. A message will be displayed on the PDA screen if the participant is to introduce the next item.

5. Technical Design
The technical design of the meeting environment (shown in figure 3) consists of a number of different levels, the human interface level, the meeting agents level, the database agents level, the database clusters and the underlying infrastructure. The human interface level controls a desktop, laptop or PDA device on which the facilitator or participant processes input and output into/from the meeting environment supporting system, displaying the information described above.

Figure 3. Meeting Environment

The meeting agents level consists of one agent for each participant and one facilitator agent. The characteristics of agents are demonstated in the following manner. Autonomy is shown because the meeting agents are the only means of getting a request from the participant to the facilitator although this autonomy is limited by the range of actions they can perform. Their social ability is exercised in recognizing the need to keep the conversation going between participant and facilitator maintaining order by

processing the requests in a timely manner. The reactivity of the agents is shown by the need to respond to requests and take action and pro-activity shown by the ability to recognise bottlenecks in facilitator response and taking the initiative to pause the request from the participant. Each participant agent links with the human interface, the facilitator agent and the other participant agents. The facilitator agent also links with the human interface to support facilitator functions. The meeting agents control the flow of messages from the participants to the facilitator. Each agent will negotiate with the other meeting agents to determine the priority of participant requests and the request with the highest priority will be actioned first. The request will be refused if the facilitator is busy processing a previous request and the participant will be sent a wait message. The meeting agents will also perform any tasks that the facilitator does not need to process personally. As the agents intercept messages on the way to the facilitator they may initiate actions automatically before passing the message on. An example of this is when the participant first logs on in which case the meeting agents will intercept the logon message and send the agenda and artifacts for the first agenda item to the participant. When the participant logs on to the system the participant agent links to the facilitator agent and sends a message to request the agenda and supporting meeting artifacts. During the meeting the participant agent processes messages from the participant and passes them to the facilitator agent. As each agenda item completes the participant agent receives a message from the facilitator agent containing the next agenda item to update the PDA screen. At any time the participant can request a copy of the agenda from the participant agent. If the participant pauses the session a message is sent to the facilitator agent and when the session is resumed the participant is updated with the latest agenda When the meeting is opened by the facilitator, the facilitator agent collaborating with the database agents controls the input of information into the agenda database and transfers the agenda to each participator agent. As the meeting progresses the facilitator agent passes messages between the participant agents and the facilitator and collaborates with the database agents to update the agenda database as agenda items are completed. Also as messages are received from participant agents they are stored on the message database with the participants name and the time received. The complete message database is then passed to the facilitator to refresh the facilitators computer screen. Certain automatic functions are performed including the timing of agenda items, the sending of messages to the facilitator when the agenda item time has expired.

279

The Database agents level will handle the storage and processing of the databases required. To prevent the problems associated with one single point of access copies of each of the database tables will be held on several computers connected to the peer-to-peer network. The database agents will negotiate with each other to provide data from the latest version of the tables in the event of failure. This process is explained in more detail later in the paper. The database clusters contain all the database tables used to support the meeting environment. These include the agenda table, artifacts table, actions table and the messages table. The agenda table contains one row for each agenda item. Each row contains the item description, the estimated time for the item, the name of the participant who is presenting the item and a status. The status will represent the condition not yet started, in progress and completed. The messages table will contain a row for each message. The row will contain message identification, message description and date/time sent. The meeting agents, database agents, devices and databases will all be connected over a peer-to-peer network. The meeting agents working for the facilitator and the participants request information from a database agent. Any one of the database agents depending on availability could respond to the request. The database agent responding will then negotiate with the other database agents to establish which agent has the latest copy of the requested information. The database agent with the latest information will then respond to the original meeting agent and then send a copy of the most up-to-date table to the other database agents. When the meeting environment is first initiated the agents will negotiate to find the latest copy of all the tables and then they will all be updated with the latest information. The same process will be repeated when the meeting environment session is terminated. The underlying infrastructure could be wired or wireless depending on the devices used by the meeting attendees. The application layer is the meeting environment which uses a peer-to-peer software solution. Each peer needs client software which creates the messages to send to the other peers and processes the messages coming back. The software on each peer would limit the scope of the artifacts for the meeting although each peer should be able to process text files as a bare minimum. The database transaction information is packed into the messages and processed by the peer client software to update the databases. The client needs to have the database management software installed to achieve this. The meeting agents can reside on the peer clients (being software processes which are initiated when the client is activated) or they could reside on other peers on the network. At the physical layer the devices will operate wired or wireless using

Ethernet protocol, each device being assigned an IP address.

6 Conclusions and Future Work


There is a need to be able to monitor the social environment in meetings otherwise the participants can become disaffected and either withdraw from making contributions, become obstructive or leave the meeting. This can happen as discussed above when the participant is unable to get into the discussion, or because a point is badly received, or if there is not a friendly atmosphere in which points can be made. It is the job of the meeting leader to ensure that participants are kept happy so that they are as productive as possible. If the participants do not make clear that they have a problem because of a reluctance to assert themselves in front of their peers or their superiors then the meeting leader will not be able to do anything about the problem. By giving the participants the means of signalling to the meeting leader unknown to any of the other participants at the meeting this situation can be prevented. The means of signalling is provided by each participant being given a PDA which communicates with the meeting leaders own computer. The meeting leader will then be kept aware of situations which need attention and be able to respond to the signals from the participants at the meeting. If the meeting leader does not respond then the meeting will be interrupted until the problem is resolved. In the near future we are hoping to conduct a user study to find out the usefulness of this service. Based on the feedback we will refine our design.

References
[1] Ellis C. and Barthelmess, P. (2003) The Neem Dram in proceeding of the 23rd conference on Diversity in computing (TAPIA 03), pp 23-29, ACM press. [2] Bulkeley, W. (1987) Better Virtual Meetings, Wall Street Journal-Eastern Edition, Vol 248, Issue 75, pB1B5. [3] Scofidio, B. (2008) Why Should(nt) You Manage Virtual Meetings, Corporate Meetings & Incentives, Vol. 27 Issue 8, p4-4, 1p. [4] Black, A&C. (2006) Getting The Best From Virtual Meetings. Bloomsbury Business Library Manage Meetings Positively, p60-71, 12p [5] Shakshuki, Elhadi; Koo, Hsiang-Hwa; Benoit, Darcy; Silver, Danie. (2008) A distributed multi-agent meeting scheduler. Journal of Computer & System Sciences, Vol. 74 Issue 2, p279-296, 18p; DOI: 10.1016/j.jcss.2007.04.019; (AN 28136281) [6] Wainer, Jacques; Ferreira, Paulo Roberto; Constantino, Everton Rufino. ( 2007) Scheduling Meetings through multi-agent negotiations, Decision Support Systems, Nov2007, Vol. 44 Issue 1, p285-297, 13p; DOI: 10.1016/j.dss.2007.03.015; (AN 26569542)

280

Kharif, O. (2007) The Virtual Meeting Room, Business Week Online,p6-6, 1p; (AN 24781227) [8] Wooldridge M., and Jennings N. R. (1995) Intelligent agent: theory and practice, Knowledge Engineering Review, 10(2), pp. 115-152. [9] Bratman M., (1987) Intentions Plans and Practical Reason. Harvard University Press, Cambridge. 1987. [10] Rao A. S. and Georgeff M. P., (1995) BDI Agents: From theory to practice, Proceedings of the First Intl. Conference on Multiagent Systems, San Francisco 1995 [11] Rao A. S., and Georgeff M. P., (1998) Decision procedures for BDI logics, Journal of Logic and Computation, 8(3), pp. 293-344. [12] Kellerer, W. (1998) "Dienstarchitekturen in der Telekommunikation Evolution, Methoden undVergleich, Technical Report TUM-LKN-TR-9801 [13] Barker, A. (2008) Crash Course in having effective meetings, Management Today, Jun2008, p22-22, 2/3p; (AN 32828858) [14] People Managenment, (2008) How to have effective meetings, Vol. 14 Issue 20, p45-45, 1p; (AN 34794091)Scofidio, B. (2008) [15] Thompson, P.; Iqbal, R.; (2007) Supporting the Social Dynamics of Meetings using Peer-to-Peer Architecture, Proceedings of the 15th International workshops on Conceptual Structure, Springer Verlag, pp, 170-176

[7]

[16] FIPA Communicative Act Library Specification, http://www.fipa.org/specs/fipa00037/SC00037J.pd f, 2000. [17] Yannis Labrou and Tim Finin, A proposal for new KQML Specifications, TR CS-97-03 Computer Science and Electrical Engineering Department, University of Maryland Baltimore County, Baltimore, MD 21250, 1997 [18] J. R. Searle. Speech Acts. Cambridge University Press, Cambridge, UK, 1969.

281

You might also like